Building a 15 W Dumble-Blues Junior Amp

Having recently built a bench version of a tube-based Dumble-like preamp that worked well I decided to built that concept into a full amp; something like a Fender Blues Junior. I wanted to build something more lightweight than a traditional 50W or 100W Dumble as it seems that musicians are getting older and don’t much care to lug around a big combo to pub gigs and the like. I had a low cost Acoustic semiconductor amp with a decent case and steel chassis that I thought was the right size and that I could modify (a lot), so I bought some parts and got to work.

This is a sort of picture book version of what I did to build the amp. I take a lot of pictures as I build something (or when I repair an amp) and I find that telling the story with pictures is way better than just saying what got done.

I began by pulling the chassis from the Acoustic combo and stripping out all the parts. The worst bit was getting the reverb tank out as it was held to the chassis with double-sided tape. A lot of prodding with a steel ruler got it loose.

When I fist got the Acoustic amp I counted up the number of pots on the front panel along with the number of LEDs and switches. The Dumble design has a lot of switches (5) and a lot of pots (7) and I also wanted to be able to include the same reverb tank and control that the Fender has. So I needed to be sure I could modify the front and back panels to include everything.

I should say that when I start out a project like this I’m usually all about “Oh, that’s going to be easy to fit all that in there.” It’s only later that we find it a bit of a squeeze as you’ll see. Since this is going to be a hand-wired amp (no PCB) I have to have room for lots of tag strips.

I took a look at the existing front panel and tried to see what it would look like with toggle switches and the LEDs I wanted to use. In particular I wanted to be sure I could fit legends around the switches. It was going to be crowded.

I wanted to use as many of the original holes as possible which meant using washers on the switches to allow me to put them in the bigger holes. This means moving the legend text further away. I also wanted to group the pots in some sort of sensible way. We’ll see how that turned out later.

The next design step was to get an idea of how I was going to fit the transformers, tubes, and several tag boards into the chassis. To do this I printed paper versions of everything so I could write on them and move them around. This is where I started to wonder if I was going to get it all in. The tag boards I use come from Tube Depot (as does a lot of other stuff).

To get an idea of the front and rear panel layouts I created some artwork in Photoshop using a millimeter grid to get the spading of all the holes as accurate as possible. I don’t have an 11 x 17 printer so I printed the test panels on two sheets of 8.5 x 11 paper and taped them together. I found out experimentally that I needed to set the zoom ratio to 98% in Photoshop when doing the prints on my HP 8100 ink-jet printer. As you’ll see I had several goes at the panel layouts. I intended the final panels to be black with white lettering but for the fit-ups it’s cheaper to print black on white.

I had originally thought I could just use tag boards for everything including the three relays and the reverb circuit. It turned out to be easier to use some Arduino Mega prototyping boards cut down a bit for these two things. The tag strips were to be used for the preamp, overdrive, power amp, high voltage supply, and low voltage supply.

Having decided on my panel layouts I taped the paper designs to the chassis and drilled out the holes. I typically center punch holes then drill with a small drill before expanding with a tapered step drill bit. These stepped drills come in US and metric diameters and are my favorite way to drill bigger holes. They do mean that you’ll end up doing a lot of deburring work, but unless you have punches or a laser cutter this is the cheap way.

I also drilled out all the holes in the base of the chassis for the tube sockets, transformer mounts and so on. Lots of holes. I made a wooden jig to hold the chassis while I drilled it. The thin steel bends too much otherwise.

I had decided to use Hammond power and output transformers, and the ones I chose were the Fender Blues Junior replacement items. They had the windings I’d need and are not too expensive. The power transformer in the Fender is mounted in a large rectangular hole but this takes up a lot of chassis space so I mounted mine on spacers below the chassis as you’ll see later.

Having drilled all the holes and deburred them, I did a fit up test of the front panel parts.

I was pretty happy with this until I tried it in the case. The position I’d chosen for my LEDs was half covered by the wood of the case. Bummer. So I re-did the front panel with the two LEDs to the left of the input jack. Here’s a picture form later on when I was doing assembly and test of the power amp.

I also had to change the toggle switches so they moved sideways rather than up and down. I was really glad I’d tried the case fit before I got any further.

The chassis got a coat of primer and them some flat black paint before I started the assembly.

As I got into the assembly (which I’ll get into in a minute) I also finalized the front and rear panel designs and went to see a local print shop about getting them printed on vinyl. The print shop offered to have their machine cut out all the holes as well which was a bonus as cutting them out by hand with an xacto knife is very tedious. The original design was done as a 300 dpi image in Photoshop. All the print shop needed was a TIFF or JPG.

The file I took to them looked like this:

I provided one dimension line just to be sure the overall length came out correct. At the time I wrote this I have not got the panels back yet.

I split the build of the amp up into phases that I could test. For no particular reason I started with the relay board which I built on some prototyping PCB cut form an Arduino Mega board.

The common Dumble design uses two relays for preamp boost (PAB) and overdrive (OD). I added a third relay to perform the Rock/Jazz function rather that using the usual DPDT panel switch. This reduces the front panel wiring a lot and meant that I could use five simple SPST toggle switches. I also modified the foot switch circuitry so that it does not need the 5-pin XLR connector, but instead uses a standard 1/4″ stereo jack. I also modified the foot switch control circuitry so that I could have LEDs in the foot switch as well as on the front panel. We”l cover all of that later.

On the relay board I used small loops of 18 AWG wire to make the wire connection points. Lots of other ways to do this, but this is cheap, easy, and works well enough. The final board was tested with a 12V bench power supply and some jumpers to simulate the switches.

In order to verify once more that I had a front panel design that woudl work, I assembled all the pots and switches with wire tails so that I had a complete assembly I could install later. A piece of bare 18 AWG wire was used to join all the pot cans together and provide a common ground for the front panel components. It’s really important not to have ground loops anywhere and this takes some planning. My chassis has a star arrangement of six solder tags which together form the single ground point for the entire amp. I also put the input resistor and coax feed line onto the input jack. This jack is insulated from the chassis by plastic washers to avoid a ground loop. The coax line is grounded at the amplifier end. This provides the best rejection of RF signals.

To make the assembly of the pots easier I did this with them on the outside of the chassis as you can see below. Once they were all wired together I did a test fit into the chassis with the switches and LEDs. This was when I found that the LEDs were hidden by the wooden case and has to do a bit of a re-design. Fortunately there was nothing much in the chassis at this point so drilling a couple of new holes was no big deal.

Next up were the high and low voltage power supplies. I built the high voltage supply first along with the mains power input and switching so that I could do a complete test of the HT voltages (albeit unloaded at this point). I like to test early and often. The screwups get found when they are cheapest to fix. And trust me, there are always a few screw ups.

Satisfied that the front panel design was now ok, I removed all the parts while I fitted the big, chunky stuff.

The original Fender Blues Junior design uses a few axial capacitors in the HT circuit. I wanted to use decoupling at each tube stage which meant a total of five high voltage electrolytics. To provide a bit os a space saving inside the chassis I used one 50 + 50 cap on the outside of the chassis and three axial caps inside. I happened to have all these capacitors in stock. Otherwise I’ve probably have used two external 50+50 can caps to save more space.

As I mentioned above I decided to mount the power transformer on some threaded spacers and put holes with rubber grommets through the chassis for the wires. The output transformer was also mounted outside with a couple more holes for its wires. Then I added the tube sockets, the can capacitor, and the tag strips for the high and low voltage supplies. Inside the chassis I built the star ground near the can cap.

I felt pretty happy with the layout in general but fiddled around a bit with which holes to bring in the power transformer wires though to keep the wiring inside the chassis tidy. The Hammond transformers have the start and ends of each winding on opposite sides of the bobbin which I find annoying. I’d much rather pairs came out together like: 110V, HT, Heater, Low voltage. But whatever, it all works in the end.

To check I wasn’t nuts with the rest of the layout I placed the rest of the tag board in place so I could see how crowded the final thing was going to be. I also did a temporary fit up of the rear panel to verify all of that was in the right places and the panel labels would be ok.

Note the use of my high voltage safety flip flops.

Yes, at this point I was indeed wondering if I could actually do all the soldering since a lot of the gubbins was very close together. ‘Gubbins’ is the term my daughter uses to describe the innards of the amps I work on. It’s an English word that means gadgets or gadgetry. Yeah – I know – not all that helpful.

Now it was time to do some real assembly. To track what I’m doing I always print out the complete schematic and then mark off the bits that I’ve wired up. This makes it much easier to avoid forgetting a resistor or capacitor. I do my schematics in KiCad (which is free) and (unlike the free version of Eagle) allows you to have multiple sheets in one schematic. So I have separate sheets for each major chunk of the amp. Here are a couple of examples printed on pairs of 8 1/2 x 11 paper and then taped together. These are the power supply and power amp sheets showing how I mark them up and record circuit modifications (which go back into KiCad).

In the first picture you can also see a drawing I made of the component layout for the tag board. I usually print two or three of these empt tag strip images and then pencil in where I think the parts will go. I use this to refine the layout so that all the parts will fit.

On a previous build of a Gibson GA-5 replica I was lucky that all the parts very easily fit between a long pair of tag strips as you can see here.

It was nice to be able to have each component between the tag strips and only wires leading to the tube sockets.

The amp in this article has a lot more parts to it and it was not possible to adopt exactly the same approach. I had to put some parts diagonally between tags and in some cases mount parts on the tube bases directly, or on the ends of pieces of wire to attach to the tube bases or pots. I hate to do that as it makes it hard to modify later if I want to change the design, but as you’ll see here it was necessary to fit everything in.

The HV power supply has some big axial caps and these fill up a lot of space, so I did a paper design of the layout and then put all the small parts on the tags first. Once those were in place I added the big electrolytics.

In order to test this piece I added temporary wiring to the mains input connector on the back panel and the power switch on the front panel. Both of these use faston connectors so it was only necessary to make up a few wires and put connectors on the power transformer input wires. Note that the picture above also includes a few parts of the low voltage power supply.

In order to test the high voltage power supply it was neccessary to terminate the 6.3 V heater (green), and 20 V auxiliary supply (brown) wires from the power transformer. It’s easy to forget stuff like that and then you have the BIG SPARKS when the power comes on :).

To test the supply I used a variac and a series lamp in the same way I test complete amps. The variac allows me to bring the voltage up slowly and the series 100 W lamp shows if we have a short in the transformer. There was no smoke and the current consumption was about zero with a good HT supply voltage on the final capacitor in the chain – so the power supply was good. The pictures below show my test bench and the variac and lamp assembly.

I built up the low voltage supply next. The power transformer (as used in the Fender Blues Junior) has a very convenient 20 VAC output which is used to provide the negative bias voltage to the power amp tubes as wel as +/- 12V for the op amps used in the reverb circuit. I’d have preferred to use a tube reverb but there was no way to fit that in here so I made a copy of the Fender circuit. More on that later, but for now I wanted to get the +/-12 V supply working so I could install the relay board.

The Fender circuit is a bit primitive so I put in 12 V linear regulators as these don’t take up much space and are easy to work with. Well, apart from the varying pin connections. It would have been so nice if whoever produced the first one (Texas Instruments maybe?) in a TO220 package had made the pins: in, ground, out. But no, the +12 V regulator (7812) and the -12 V regulator (7912) have different pin connections, so be careful. I got this wrong the first time and had to spend an hour figuring out what was wrong. Today’s moral: don’t guess. Look at the data sheet for this exact part.

I built the low voltage supply in two stages. First I put in the rectifier diodes, a resistor, and some capacitors. I wanted to test that first before adding the regulators. On power up I watched the positive and negative voltages on my home built multi-channel voltmeter and wondered why the -12 V supply was zero until the small electrolytic capacitor went ‘pop’ and I realized I’d put it in backwards. It’s been a few years since I’ve done that so I’m guessing the senility is progressing.

Once the capacitor was fixed I added the regulators and spent a happy hour or so sorting out the incorrect pin wiring. I’m not usually this brainless. It was just one of those days. The picture below is during the mystery pins phase.

Another note about testing: While working on the low voltage supply and getting that working, it’s really important not to forget that the high voltage supply is also on, and right there ready to be touched. I use a discharge wand with a beefy 1 kohm resistor inside to discharge the big caps every time I turn the power off.

Once I had the power supplies working I decided not to install the relay board. I thought it would be better to get the power amp working. All the preamp and relay stuff had been implemented on a bench model of the preamp a while ago to debug the schematic I’d created. A picture of it is below:

The bench model uses an external power supply and power amp. This was just to get the complete Dumble preamp debugged.’

Before we get into the power amp build which is the next major chunk, I need to sidetrack a bit and talk about the reverb circuit. The Fender design uses a spring tank which is still available from Tube Depot and others. The specific tank has a relatively high impedance input driver so that it can be driven by a 4560 op amp. Most tanks in older tube amps have low impedance inputs and are driven by a tube stage that uses a 1 W audio output transformer.

I wanted to reproduce the Fender design (I can hear you asking why) so I built up the 4560 circuit on a breadboard on the bench and connected it to a new Accutronics tank I’d just bought. It didn’t work at all. Looking at the scope I could see almost no drive signal at the tank input. Lots of debugging later I found that my circuit was just fine but the tank wasn’t. The tank I’d bought new from Tube Depot was marked correctly but was evidently the wrong innards. I’m guessing that a batch at the plant in Korea where these are made got the wrong labels stuck on them.

To make sure I wasn’t imagining things I measured both the DC resistance of the input and output transducers as well as the AC impedance at 440 Hz using a signal generator and a scope. This was definitely the wrong part.

The folks at Tube Depot have been incredibly helpful to me in the past when there has been anything at all wrong with any of my orders (which is rare). So I wrote up a nice email with my readings and some pictures of the tank and asked for a replacement as well as suggesting they test any others in stock so that they don’t get sent out to other Blues Junior repairers.

I got back a nice email telling me that the DC resistance and AC impedance are not the same and the tank is fine.

Hmmmm. I politely replied that I have a degree in electronics, years as an electronics engineer, spend hours repairing tube amps and had written down all the evidence in me original email. I’m guessing the email was only scanned, not read in full – it was quite long.

I got a really nice response and a new tank in the mail. I did a quick DC resistance test when it arrived and it seemed fine. At the time I’m writing this I have not done the impedance test at AC or tried it on the bench. Later for that.

Back to the power amp. The power amp in this design uses two EL84 tubes just like the Fender Blues junior and a few other amps. My implementation of the phase splitter and power output stage is an exact copy of the Fender circuit. The only difference is that I decided it would be nice to have variable bias on the output tubes rather than the fixed -10.7 V in the Fender design. So I looked up the specs for the JJ EL84 tubes I was using and decided that a voltage range of about -8 to -12V should be good. This was great as I already had a -12 V regulated supply in the design. All I needed was a pot and resistor to set the range. Lovely.

I built up the heater wiring for all the tubes sockets (very tedious work with 18 GA stranded green wire) and then added the few resistors and output transformer connections so I could power up the tubes and set the bias.

Once again, I used the variac to bring up the AC voltage slowly while looking at the DC HT voltage, the bias voltage on the grids and the current through the cathodes (by measuring the voltage across the 1 ohm cathode resistors). When the input got to about 100 VAC the bias current with -12V on the grids was already up to 25 mA. At 110 VAC input it was over 40 mA. I switched it all off. The cathode current for mid-range bias of an EL84 at 350 V on the anode in class Ab is about 20 mA. So clearly I needed a lower (more negative) bias voltage. I also wondered if I’d misread the Fender circuit.

It turns out that the Fender Blues Juniors run hot, and I didn’t want that so I modified my bias supply circuit so that I could get as much as -20 V on the grids. Now it worked as I expected and I had a good bias current range I could select with the pot I’d installed on the chassis underside.

I should note here that this amp is sort of a prototype and as such I have pots exposed for a few things that would usually be hidden away from interested fingers. “What does this do?” is not a question I want to hear from someone altering the output stage bias. So my amp has a couple of “Danger Will Robinson” items that would be removed or hidden inside on a more consumer-oriented version. You’ll see the other two later. These are Dumble concepts for altering the overdrive input level and modifying the phase splitter balance.

Here are pictures of the output tube wiring and the test setup for measuring the bias.

As you can see I have several components mounted around the tube bases which I normally try to avoid. This design has a lot of gubbins in the power amp. Here is a rough layout of the tag board I did for the complete phase splitter and power tube section. I think it took me two goes to get all the parts on the tag strip once I’d decided to put a few on the tube bases.

Wiring up the phase splitter and power tube circuit for the power amp took a couple of hours. It was really helpful to have the rough tag strip layout on the schematic as I worked. Once that was done I added tails for the 8 ohm speaker that would be used later on. I needed these connections so that I could run a signal through the completed power amp. I connected the speaker lines to my 8 ohm dummy load and did a final visual check of the phase splitter tube socket before powering up.

I did the initial power up with the variac until I was certain that we had no smoke. I immediately found one missing solder joint as one of the phase splitter anodes had no voltage on it. With that corrected I ran a signal through the power amp. I’m going to point out again here that before making any changes the procedure (for me anyway) is to switch off the amp, switch off my variac outlet, and then discharge the big high voltage caps before proceeding.

I do measurements whenever I power up a new stage. Once the wiring was right I measured:
Mains in: 0.28 A
HT: 350 V
Grid bias: -17V
Bias current: 19 mA

With a 440 Hz signal driving the input to the phase splitter I looked at the output waveform into the dummy load using my scope. With the input level wound up it clipped at about 38 V peak-to-peak which is about 23 W into an 8 ohm load. That all seemed good, but it was evident that I had some mains 60 Hz ripple on the output.

I’ll skip the investigation other than to say that I looked around with the scope until I found the stupidly small capacitors on the inputs to the low voltage power supply. In some mindless moment I’d put in some very tiny electrolytics instead of what was really wanted. Having added in a couple of 1,000 uF 50 V caps it all looked much better on the scope. I also added a small capacitor across the 20 V Zener diode I was using to keep the bias voltage stable. Probably not necessary but it pays to keep noise out of the bias chain.

The last picture shows the meter and scope probes in place as I did the initial testing. Here is part of the power supply schematic showing the final low voltage parts.

And here is the complete power amp schematic. Bear in mind that I might tweak this later. The full schematics will end up at the very bottom of this article when the amp is finished.

Since I’ve decided to work backwards from the output stage, the next thing to get implemented will be the overdrive stage. This uses two triodes and is based on the published Dumble-like preamp designs. This is the same as the overdrive in my bench model of the preamp.

Before getting on with wiring up the overdrive section I decided to test my replacement reverb tank. This one checked out great on the bench and worked well with my breadboarded reverb circuit. That’ll all show up in the build a bit later. For now, let’s see the overdrive build.

The Dumble overdrive uses two triodes (in an ECC83/12AX7) so it has two gain stages and will easily square off any signal presented to it. Dumble seemed to like twiddly bits in his amp mods so the overdrive circuit uses three pots: Trim, Drive, and Ratio. The Trim pot is normally inside the chassis as a preset but I put mine on the back as this is a prototype and I wanted to be able to fiddle with it easily during testing.

As with other parts, I did a paper layout of the components on the tag strips.

Once I was happy with the layout (which I modified only slightly as I got into the implementation) I found all the parts and laid them out on the schematic. Then the bar tag strip was screwed into the chassis and I was ready to begin.

For this section of the build I decided to put all the parts on the tag strip before doing any wiring. I just wanted to see how I felt about doing it this way. I think it’s better to add parts and wires as you go, since this makes it easier to mark what you’ve done on the schematic. In this instance I added all the parts to the tag strip and then marked things off as I wired it up. I thought this was more error prone so won’t do it this way again.

Here is the assembled tag strip in the chassis.

As you can see there was plenty of room so it turned out to be a nice layout. the 68k and 220k resistors at the bottom later got moved to the ends of the wires going to the Trim pot as this kept the wiring shorter.

The wires to the pots got added next along with the B+ supply and a ground. Here’s the board wired up in place.

This section can’t be tested easily without adding in the relay board so that comes next.

The relay board was built as one of the first jobs I did. Here is the layout sketch I did and the board after it was assembled and tested.

This would be oh so much nicer on a small PCB and I did consider that. Prototype PCBs run at about $5 per square inch. This board is 1.5″ x 4″ so that would be $30 for the PCB. So not horribly expensive but at this stage I’m not sure what I want on a PCB in the final version.

As I was getting ready to test the overdrive section, I found out that my panel prints were ready so I went off to pick them up. I’d produced the panel images in Photoshop – not the ideal tool but does everything you need including placement grids, accurate font sizing and so on. Getting a panel aligned with the existing holes was the worst part. Sure it’s just a case of measuring the hole positions and transferring that to the image but with an existing chassis that has curved top and bottom edges, it takes some fiddling to get accurate baseline references. OK, so I’m doing some whining.

To test the panel layouts I printed them as black on white (to save ink) on plain paper. By experimenting with my HP printer I found that I had to set the print scale to 98% to get good results. I’d added a dimension line to the artwork so I could verify it easily. The awkward bit is that I don’t have an 11 x 17 printer – just 8.5 x 11, so I had the print the panels on two pages and join them on the table.

The printer’s machine had printed my artwork on sticky vinyl and cut out the holes for the pots, switches and jacks. I was a bit nervous about getting the sticky plastic lined up on the chassis as the chassis was a bit awkward to work with now that it had the transformers mounted on the outside.

I did the rear panel first and that came out OK, so I was happier about committing to the front panel.

I cut the printed sheets down to the size of the metal panels so I could tape one edge with blue painters tape before pulling off the backing paper and sticking it down. The vinyl I chose is very thin and stretches a bit if pulled so I had to be careful to get it down flat but not pull on it too much.

The printing is a little fuzzy in places. I used a 300 dpi image so I doubt it’s that. If I do this again I’ll use thicker vinyl and a larger font perhaps. Overall I thought the result was pretty good for a custom-built prototype.

Happy with my new panel I went back to testing. I started out verifying that the overdrive (OD) and preamp boost (PAB) switches worked the relays as well as the rock/jazz switch and its relay. I also tested the foot switch input with my foot switch test box and then the actual foot switch I’d bought for this amp. Once all that was done I put some signals into the overdrive stage and verified that it worked.

All that is left to do now is implement the preamp and the reverb. The preamp has a lot of components in it thanks to Mr Dumble’s predilection for twiddling. I’m hoping that it will all fit on the one remaining tag board. So next step is to print some tag board work sheets and try to layouts.

I had been a bit concerned that the preamp wasn’t going to fit on the tag board I had chosen for it. So when I’d installed the tone control pots I’d added a few components directly on the pots. I also decided to mount the input resistor on the jack (common practice) and put one more resistor inline with the coax from the input jack (also common practice). I would have much preferred to have all the components on the tag board as it’s just easier to test like that and much easier to swap them if I need to. If I do a PCB for this amp then everything possible will be on the PCB with just jumper wires to the pots, jacks, and tube sockets.

As for the other tag boards I did some pencil layouts to get an idea of where all the parts would go. I had multiple attempts at this one to try to keep it easy to wire up. The pictures below sort of show the process.

As I wired this tag strip into the amp I marked off the wires and components as I went. The last picture shows it all connected and the next picture shows what it looks like in the chassis.

The open space is for the reverb board which I’ll get on to next.

As the preamp is tied to the relay board and I wanted to be able to test it fully I decided that now would be a good time to modify the foot switch I’d bought from Tube Depot.

I’ve altered the foot switch circuit from what Dumble did for two reasons. First, I wanted to get rid of the 5-pin XLR he used on his amps as those are expensive. The second reason is that I wanted a normal two-button foot switch with a 1/4″ jack to work if it was plugged in. So I re-designed the relay drivers to use transistor switches. This allowed me to have LEDs in the foot switch box but still only have a 3-wire connection. The transistor switches on the relay coils also help localize the relay switching current and hence reduce pops in the amp as you alter the OD or PAB selections.

The following pictures show part of the relay driver circuit and the wiring required for the foot switch itself.

To add the relays to the foot switch I pulled it apart and then drilled to 1/4″ holes for the LEDs. They got wired up and the switch reassembled. I did a quick bench test before closing up the switch case.

The final part of the chassis build was to assemble the reverb board and wire it into the chassis. As I think I mentioned earlier, when I got into this project it all started to look a bit crowded and I wondered if having the reverb was worth it. But I’d bought the reverb tank and foot switch so I decided to squeeze it all in.

As with the tag boards, I did a paper layout for the reverb prototype board. Once I was happy with that (two attempts) I wired it up.

To test the reverb I added some temporary power and signal wires to the board and made up the audio cables which were also attached temporarily. The whole assembly was then wired up on my prot board as that has power and signal connections available. The board tested fine, so all the temporary connections were removed so it could be installed in the chassis.

Finding audio cables with RCA jacks that are of good quality is a bit of a crap shoot. The first cables I bought had no screens and the wires inside were plain copper (not tinned) so destined for some corrosion later on. The ones shown here were pretty good. They have a foil screen with ground wire. The conductor wires were again plain copper but I tinned the ends all the way back to the sleeving and I think they’ll be ok. By the way, having a paper sketch of the layout of the board and the location of the connection points is essential for assembly into the chassis and subsequent testing.

The reverb board was then fitted into the chassis and wired up. I’m really pretty unhappy with how crowded things are. It makes it difficult to do nice, tidy wiring. I will say though that running lots of audio wires in a tight bundle is asking for trouble as the close proximity increases coupling capacitance. Despite looking untidy, having wires that cross each other rather than running parallel is better electrically.

With the final assembly done it was time to walk away for a minute before firing it up for a test. I like to be methodical about testing, and being all fired up with excitement isn’t the best way to get started. Yeah, I know – pretty boring, but I like to be sure I have the amp set up correctly (pots mid way, etc) and the right signal connections in place before the power goes on. In this case I’d powered it up lots of times by now so I wasn’t worried about smoke.

With the reverb tank connected I fired it up and discovered that it all seems to work correctly. So after the bench test it got reassembled into the case from the original Acoustic amp I’d cannibalized to make this Dumble-Blues Junior hybrid.

Before putting the chassis back into the case I wanted to take a few more pictures. The chassis had been upside down for all of the build. When I turned it over (see next picture) I realized that the wood block I’d been using as a leg was still screwed to the chassis. And …. the wood screw holding it place was under the reverb board. So that was today’s screw-up. I undid the screws holding in the reverb board and removed the woodscrew. Now I could take the top-side pictures and assemble the cabinet.

I mounted the reverb tank in the bottom of the cab with woodscrews, washers, and plastic grommets. Fortunately cabinet material was soft enough to get the screws in by hand.

I had planned to replace the original 12″ Acoustic speaker as part of the build but all I had on hand was a 10″ Jensen Mod 10/35 that I had bought for another project, so I thought I’d test it as-is and replace the speaker later if it sounded awful. Bear in mind that this amp is strictly a prototype. If it works well and the features turn out to be what’s wanted then I’ll do a PCB design so that a Fender Blue Junior can be retrofitted with this Dumble-style preamp and the Fender reverb.

The front panel is a bit of a squeeze in the cabinet hole but acceptable for testing. I was almost tempted to make a new case but that’s a lot more work just for a prototype.

The next step is to take the amp to Ruse Kruse at Florida Tube Amp and get a no-bullshit assessment. Stay tuned …

Fast forward a day. Here’s a short clip of Russ playing through the amp.

Overall the amp did quite well, even with its crappy original speaker (from the cheap Acoustic amp I used for the case and chassis), but it clearly needs a better speaker. Here is the list of things that need to be changed or fixed (in no particular order):
* Add a jack in the back for an external speaker
* Put in a better speaker
* Bias the EL84 power tubes a bit hotter
* Fix the hum that mysteriously shows up in overdrive mode
* Add some top-end cut for the reverb

After some discussion with Russ and a bit or digging online I decided to go with a Celestion Vintage 30 speaker, and ordered one the same day. I’d considered putting a jack on the bottom of the amp for the speaker connection when I built it but there was precious little room given all the circuitry I had to get into the chassis so I went for a direct connection. On reflection it would have been pretty easy to add the jack. I didn’t want to drill a new hole now that the amp was built so I decided to add a small bracket and jack in the back of the cabinet.

When I initially set the amp up I biased the EL84s a bit on the cold side. During testing with a since wave signal and looking at the output on a scope I could see a hint of crossover distortion at full power. Since this amp is likely to be used with the overdrive on most of the time, a little crossover effect is down in the weeds. But it’s easy to alter the bias as I put a pot through the base of the chassis. I will say that you don’t generally want to just twiddle a bias setting unless you can also monitor the bias current through the tubes as it’s easy to get the tubes running very hot – way past what they are rated for. That’s going to radically shorten the tube life. This amp has 1 ohm resistors in the EL84 cathode leads so I can measure the bias easily when the amp is on the bench. With it in the case requires using a bias probe adapter on one of the tubes. Russ’s suggestion was that I bias at about 75% of the maximum power load for the tubes. The plate voltage is about 350 V on the EL84s and they are rated at 12 W maximum dissipation, so 75% of that means 9 W on the tube. With 350 V on the anode, that’s a current of 26 mA. I’d initially set the bias at about 20 mA, so the increase isn’t that great.

Russ thought the reverb could do with a little lower high-end. The tank in this amp is the same Accutronics tank used in the Fender Blues Junior. I deliberately chose this tank as the aim of this project is to produce a PCB that will drop into a Blues Junior to modify it with the Dumble features. In the Blues Junior the tank is driven by a 4560 dual op-amp IC. I wasn’t keen on having a semiconductor driver in what was otherwise a pure tube amp but since this was how the Blues Junior is built I decided to go with it rather than adding another tube and transformer.

I prototyped the reverb circuit on the bench and found that the original design is a bit prone to high frequency oscillation (depending on the circuit layout) so I modified the Fender design slightly to add a high frequency roll off at about 7 kHz in the driver stage. Here’s the reverb circuit exactly as I built it for the prototype.

The only change from the Fender circuit is the addition of C708 in the driver stage. I did also change R707 from 2k0 to 2k2 just so that it was an E12 value that I had in stock. So to lower the high frequencies in the output of the reverb we need to make C705 bigger. Right now the roll off frequency is around 15.4 kHz – which is really very high. Most of us old farts can’t hear much beyond 5 kHz. So I’ll do a little experimentation and find a bigger value for the capacitor to take the edge off a bit. This is the kind of subjective stuff I really don’t like. I’d rather someone say “You need to roll the reverb off at 4 kHz” (or whatever), so that I can calculate the right value. Oh well, some tinkering will be required.

The most annoying result of the testing was the hum we heard when the overdrive was selected. I put a lot of effort into wiring any amp so that there are no ground loops and so that every stage has adequate high voltage power supply decoupling. This amp has a bit more decoupling in the preamp supplies than the Blues Junior design and I used a star ground so that each chunk of circuitry has its own ground wire.

Hum can get into a tube amp in lots of ways: bad heater wiring, insufficient decoupling of the bias circuit, ground loops, poor B+ decoupling, and more. Finding the source usually requires some time with the scope looking at possible source locations. I’ll let you know how that goes. The biggest annoyance for me was that I didn’t notice the hum days ago when I tested the overdrive with the amp on the bench. It’s really unlikely that the aliens added some hum when I put the chassis in the case. It’s mostly likely that I wasn’t listening for hum – just the overdrive effect.

So that was all a lot of chat about a few small changes. Imagine what it’s like to design an amp from scratch!

I forgot to mention that once I had the full amp built I ran a quick frequency plot with the amp in clean mode (no overdrive, preamp boost off). Here’s the plot showing the range of the bass, mid, and treble controls.

I ran the plot mostly to see that the tone control stack was working. This plot was done from signal input to the speaker output signal into a dummy load, so it ignores the effect of the speaker. Once I have the Celestion speaker in the amp I’ll do another plot using a reference microphone in front of the speaker so we can see the true overall response.

Stay tuned …