1/31

Hi Brian

The amp arrived OK today and I got the power supply built and running. With
123 VAC input I get exactly 300 VDC with a 1250 ohm load (240 MADC). This is
also with the four 2A3's plugged in so as to draw the additional 25 W of
filament power (the filament voltage read 2.45-2.47 VAC---very nice). It
looks like it should be fine for the class A amp.

I have included two photos of a modification of the AC input circuit. Since
we are either paralleling or putting in series the two windings it would be
helpful to have a terminal strip to do these connections more easily. I
added one and you will see it off to the right of the image.

I also noticed some minor mistakes in the manual that I am sure you are
aware of. The primary wiring diagrams should read 0 110V 123V not 0 110V
220V. Also on page 5 at the bottom it should read position 1 not position 3.

This is a really nice product, Brian, and the power transformer appears to
be a beauty. I will reserve judgement on the outputs until I get to that
point!

1/31

As I start to modify the Joplin I did notice that the bottom cover is highly louvered. While this is a good idea I am afraid that they won't do much good. There is no air route on the top of the chassis to allow convection to occur. I don't really think heat is a major problem with this amplifier but the internal chassis temperature would be considerably lower if there were eight holes around each 2A3 socket. These could be an 1/8" in diameter. This would facilitate air flow from the chassis while at the same time providing convection air cooling for the 2A3's.

I won't attempt to make these holes on mine as they really need to be punched to look good.

Regards,
Lance

2/1

Hi Brian,

Before I get into where we are I found two generalized items you may want to add:

1) I assembled the tube cage to make sure that it didn't interfere with the new added parts. I noticed that the top cover holes for the cap screws are a little large. This is a help when assembling everything as it allows a bit of "slop" to make sure that everything lines up right. The problem is that the cap screw tends to partially go into the oversized hole when tightened down. I suggest adding a black flat washer here. That would allow the slightly oversized hole while still looking good.

2) I removed the wooden front panel when I was drilling holes in the chassis for obvious reasons. I reattached it before adding the new parts as the screws for the front panel would have been a little tough to reach. As you can see in today's photos I added a strip of adhesive tape around the edge of the panel to protect it during assembly. You may want to add this to the assembly instructions.

BIAS MODS - first steps

You will see in the various photos that I have installed the new bias rheostats and bias point tip jacks. I think it looks pretty good and there is room to put everything in. I tried to keep hole drilling to a minimum. You will notice the four red tip jacks, one next each 2A3 socket, that will be used to check bias. There are also two black tip jacks which are for the meter common. I used two for symmetry and convenience reasons but I am sure you could get by with one for production. Under the chassis you will notice a vertical terminal strip on each 2A3 socket using the original hardware. Also, on two of the mounting screws for each output transformer you will notice one lug terminal strips. There was room for everything.

I had to trim the corners of the main PCB to clear the black tip jacks but you could move them, or it if you use one, to clear the PCB. I was being a little conservative as I wanted to be sure and clear the 2A3 bulbs and the transformer cans.

Here are the parts I added:

4 100-ohm 4 watt rheostats
4 red tip jacks
2 black tip jacks
4 one lug terminal strips
4 two lug vertical terminal strips

Holes drilled (you would punch these in production obviously):

4 3/8"
4 1/4"

I didn't bother with part numbers as I am sure you will use Chinese parts. Finally, things are a bit tight but there is sufficient room for the Alps pot without squeezing it in. It will need to be wired out of the chassis and then installed into the hole wired as there will not be enough room for in-chassis assembly. I think this is easier anyway.

I retook the AC primary terminal strip photo as I saw that it was a bit fuzzy (I had the macro off). I have attached it.

That's it for now. Hopefully in the next day or so I'll get it up and running.
Regards,
Lance

2/2

Hi Brian,

Well, I've started to wire up the bias circuitry. I wanted to verify operation so I built the bias circuit onto one 2A3 only. I then clip leaded into the power supply for DC tests only, no input and no output circuits. I also hung a 1700 ohm 50 W resistor across the HV supply to simulate the load of the other three 2A3's. Everything fired right up and I observed an adjustable range of 53-63 MADC. At 123 VAC input the voltage on the plate of the 2A3 was 299 VDC. Common to filament resistive center tap was 44 VDC (nearly the data sheet bias point of -43 VDC). This put the voltage across the tube at about 255 VDC....right where it should be. The bias adjustment is smooth and completely stable.

Each side of the ouput transformer has a DC resistance of 95 ohms. This will be in series with the 2A3 and will drop the plate voltage about another 5 VDC. This will bring things perfectly to 250 VDC.

You will see in the photo the re-placed original one ohm resistor which goes between the DC common lead and the tip jack (the bias voltage sense point...@ 60 MADC a voltmeter will read 0.06 VDC). This is in series with the 100 ohm 4 watt rheostat which is connected up as a varible resistor. In series with this is a 360 ohm 5W resistor that is in turn in series with a 400 ohm 10W resistor. Why the mixture of power ratings? It 's done for mechanical reasons. The price of a 10W resistor is only pennies more than a 5W resistor in large quantities. The extra length of its body saves another terminal strip and short piece of wire plus the labor to wire it up. I know that I have two different types of resistors, a ceramic 5W and a vitreous enamel 10 W, but it was all I could find on short notice. Normally, I use vitreous enamel power resistors because they look nice. There is no difference between the performance of vitreous enamel and ceramic resistors. Vitreous enamel does cost more though. A single 10W resistor could be used as the total dissipation is about 2.6W. While this doesn't seem like much in a 10W resistor they will get quite "toasty" and become a fairly good spot heat source. Dividing the resistor in two spreads this heat source out. This will become important after years of use. On the original hum pot I added a 27 ohm 2W resistor to each leg. This makes the AC hum adjustment a little smoother. You can also see the large electrolytic which is the bypass capacitor (220 uf@160VDC).

I will now wire up the rest of the bias circuits. By the way, the Alps pot clears everything nicely.

Regards,
Lance

2/3

Brian,

I have now finished the Joplin Class A modification and I must say that I am really satisfied with the results. Let's take up where we last left off.

Because of the bias circuitry that must be added to each 2A3 the Alps pot and cable assemblies must be completed out of the amplifier. Even if you don't do this mod I think that it is easier to wire up the pot and complete the cables out of the amplifier. I have attached a phot of whato the completed pot/cable assemblies looks like.

Finishing the amplifier, after the bias circuitry has been installed is pretty much like the instructrions for the original model. I did a few things differently and I will run through them now.

1) The main AC primary cable harness no longer circles around the right edge of the chassis as shown in the original instruction manual. I have included multiple photos of the new routing scheme that I used to allow room for the bias circuits.

2) Pay special attention to the connections to the speaker common terminals. There are now four on each one. They are as follows:

OPT "0" lead
Bias circuit common 2A3 #1
Bias circuit common 2A3 #2
Connection to "ground first" land on the main PCB.

This repeats for the other side.

3) Please note the two black wires that connect to the main PCB lands 12/13 and 16/17. Since we are now using cathode bias the cold end of the 100K 2A3 grid resistors need to be returned to common. These two black wires, visible in the photo, connect 12/13 and 16/17 to a PCB via that connects to "GND END".

4) The bias common tip jacks (black) connect to 12 and 17 on the main PCB.

5) I tie wrapped the input shielded cables to some of the bias circuit black common wires for 2A3 #1 and #2 as shown in the photo. This is more stable mechanically.

6) I added the wires that go to the feedback resistor to the speaker connectors even though I had removed both resistors. This will allow one to play with feedback later if so desired.

After the wiring was completed I turned the unit on for the first time. All the 2A3's biased right up except tube # 2. It appeared dead. I had pinched the wire for the bias circuit connection between the 100 ohm pot and the tip jack when I installed the Alps pot. After replacing this wire all tubes biased normally. The hum balance adjustments were smooth and straight forward. The bias adjustments are gradual and smooth with none of the "jumpiness" that some have noted on the class AB1 Joplin.

When I attempted to inject signal into the amp the right channel input showed 6 VAC at the RCA input jack. The 12AT7 was not lit. Some quick continuity checks showed that a mistake was made during the assembly of the main PCB at the factory and a heater connection was made to pin 1 instead of pin 9. I removed the main PCB and put the under board heater wire to the right pin. After everything was back together I fired the amp up.

There was instant signal out of both sides. It works!

My next email will give test results. I have also attached a photo of the completed amp bottom.

Regards,

Lance

2/3

Now that the amp is working I wanted to give some test results, and listening impressions. Before I start though here are the parts that go into each bias circuit:

(1) 100 ohm 4W rheostat
(1) 360 ohm 5W resistor
(1) 400 ohm 10W resistor
(2) 27 ohm 2 watt resistors
(1) 1 ohm 2W resistor (part of the original parts package)
(1) 220 uf/160 VDC electrolytic
(1) 2 lug vertical terminal strip
(1) 1 lug terminal strip
(1) red tip jack

(1) black tip jack per pair of 2A3's


Static DC parameters @ Vin AC = 123 VAC

Common to Filament resistive center tap midpoint = 45 VDC

Voltage from 2A3 plate to common = 290 VDC

Voltage across the 2A3 (filament to plate) = 245 VDC

Plate current = 59 MADC


I ran power output and distortion checks for each channel, here is what I got:

No feedback, Load = 8 ohms

Right Channel

@ Po = 1W

20 Hz, Vin = 0.07 VAC, Gain= 32.1 dB, THD = 0.3%
1 KHz, Vin = 0.53 VAC, Gain = 34.5 dB, THD = 0.22%
20 KHz, Vin = 0.06 VAC, Gain = 33.4 dB, THD = 0.17%

Gain flatness = +/- 1.2 dB, 20 Hz - 20 KHz

@ THD = 5%

20 KHz, Vin = 0.18 VAC, Gain = 32.2 dB, Po = 6.7 W
1 KHz, Vin = 0.17 VAC, Gain = 34.1 dB, Po = 9.2 W
20 KHz, Vin = 0.2 VAC, gain = 32.4 dB, Po = 8.6 W

Left Channel

@ Po = 1W

20 Hz, Vin = 0.056 VAC, gain = 34.0 dB, THD = 0.6 %
1 KHz, Vin = 0.047 VAC, Gain = 35.6 dB, THD = 0.56%
20 KHz, Vin = 0.053 VAC, Gain = 34.5 dB, THD = 0.28%

Gain flatness = +/- 0.8 dB, 20 Hz -20 KHz

@ THD = 5%

20 Hz, Vin = 0.13 VAC, Gain = 33.4 dB, Po = 4.7 W
1 KHz, Vin = 0.14 VAC, Gain = 35.2 dB, Po = 8.2 W
20 KHz, Vin = 0.16 VAC, Gain = 34.4 dB, Po = 8.8 W

I noticed that at 20 Hz the OPT's were starting to "run out" of iron a bit. I raised the test frequency to 35 Hz and things improved quite a bit including a much better looking waveform at high power:

Right Channel

35 Hz, Vin = 0.16 VAC, Gain = 33.4 dB, Po = 7 W, THD = 2.25%

Left Channel

35 Hz, Vin = 0.0.135 VAC, Gain = 34.7 dB, Po = 6.7 W, THD = 5%

Also note that I made absolutely no attempt to select tubes for distortion or gain matching. I just plopped in what I had. Even though the initial results are really quite good, an hour spent swapping tubes around would result in almost perfect gain balance between the two sides and better distortion on the left channel (my experience with 2A3's tells me that distortion slope I saw is due to an "ugly" 2A3.)

The AC hum is really non-existant. This is a very quiet amplifier. (hats off to your designer, Brian. Tell him, from one professional to another, that he did a marvelous job of implementing the grounding circuitry).

Listening tests were next. Even though nothing is broken in the sound was transparent and light as one would expect from 2A3's. The added punch from another tube on both sides (versus 2A3 SET) makes this amplifier more versatile with one's choice of program material. Things will only improve with breakin.

Finally, I have included a photo of the finished amp. Class A is the way to go with the Joplin.

Regards,
Lance

2/5

Brian,

I am doing one last item on the class A modification. If you will remember I had the cathode resistor as a 360 ohm @ 5w ceramic and a 400 ohm @ 10 w vitreous enamel in series. I have now ordered the proper resistors so that they will look good and distribute the heat more equitably (the 5w would run a little hotter than the 10w because of the values I had on hand).

The new 5w resistor value will be 250 ohms and the new 10w resistor value will be 500 ohms. They will also look the same.

Looking at the hum balance pot solder lugs I do not think that the center lug is husky enough to support a single 20-25W resistor. If one wanted to use a single resistor of this size then another terminal strip (1 lug) would need to be added.

I will send new photos (this weekend) when I install the new resistors.


Regards,
Lance

2/8

Brian,

As I said earlier this week, I have installed the "proper" resistors in the class A Joplin. The 5W and 10W resistors in the bias circuit are now vitreous enamel Ohmites. They are very rugged and look pretty good. I have attached a photo of this modification.

I also have been playing with feedback. Normally, I don't care for feedback with SET. This particular circuit seems to benefit a bit from a slight amount. With the stock 20K resistor the measured feedback was 2.8 dB. This is not very much but it did incrementally lower both THD and improve gain flatness. As with most global feedback schemes the amplifier is slightly less sensitive to different tubes. Listening tests indicate a bit more tautness that is pleasant with some program material. For the time being I've left it in but it's certainly a personal choice.

I have one other experiment I want to look at. The cathode resistor bypass capacitors are 220u/160V Rubycon electrolytics. usually, this capacitor does not affect the sound as much as say a coupling cap but I do want to try a better quality cap here nevertheless. The one I'm looking at is the 82 uf/250V Solen. These are big and expensive ($25-30 each). One change that will have to be made before though is that the hum balance pots will need to be changed to Clarostat 4W wirewounds which are mechanically more rugged. The center wiper lug on the present pot will never support the mass of the Solen cap (45 x 58 mm). After I get a few more hours of break in on the amp I'll make these changes.

I now am doing some pretty exhaustive listening tests as the amplifier breaks in. I like what I am hearing more and more and both imaging and definition are superb. The amp only has about 10 hours on it and I am certain as the hours roll by it will sound better and better.

Sometime in the future I'll boot out the Richardson 2A3's I have know (remarked Sino's) and put in the single plate Sovtek's. This is a tube I am very fond of.

I'll keep you posted.

Regards,
Lance
Brian,

I forgot to add something to my last email.

I also changed the values of the two Ohmite power resistors slighty to equalize the heat dissipation between them. The original mode had the 5W resistor running slightly hotter than the 10W. While this is not a huge problem since the dissipation is 2.6W between the two of them but I did want to spread the heat out a bit.

The original total resistance was a 360 ohm 5W in series with a 400 ohm 10W. The new string is a 250 ohm 5W in series with a 500 ohm 10W.

I know that this is probably hair splitting to some but I have found that spreading the heat out does have an effect on lowering the level of "cooked" components if looked at in terms of years of operation.

Please feel free to add this email and the last one to the web page on the class A mod.

Regards,
lance

Phase 3 - 47

2/18

After doing the 2A3 Class A mod a few weeks ago I continued to search for further improvements that one could make to the Joplin. During an exchange with KYW and Ivan303 a discussion of vintage tubes came up. KYW tried 45's in the Joplin and said that he liked them very much. Good 45's are now a bit pricey, if available at all, and Ivan303 suggested the 46. Normally i don't like to use vintage tubes as they are getting increasingly hard to get.I became curious; however, and tried some 46's in my SET test jig as I had a few on hand. I was not impressed as I said in an earlier post. The sound was nondescript and distortion, from tube to tube, was all over the map. Further searching led me to think that the 47 (directly heated pentode) might be interesting in a triode connection. I also had some of these and the results in my test jig were much better than the 46 (the 47 has a different grid structure but has the same filament and plate).
After success at the test jig level I decided to convert on-half of the Joplin to use 2 X 47 ST (in this case National Union). I swapped out the tube sockets and installed two 5 pin ceramic plate sockets. I jettisoned the Rubycon 220uf/160V cathode bypass caps I used in the Class A 2A3 mod and I put in Sprague Atom 100 uf/100V caps as they are slightly better quality although I believe the change has minimal effect on sound quality. The cathode bias resistor string became simpler and is now just the 500 ohm 10W in series with the 100 ohm rheostat on each tube. The 47 screen is connected to the plate via an 82 ohm resistor. This completed the physical mod.
I experimented with the bias point and selected one that has a plate dissipation of 11 W (the rating of the 47). I wanted to suck a little more current through the tube to better match the Joplin's 5K P-P output transformers. My final bias point is:
Voltage from plate to common = 298 VDC
Voltage from filament to common = 21 VDC
Voltage across the tube = 278 VDC
Plate current = 39.5 MADC
The performance was staggering as indicated below (with 2.8 dB global feedback):
Right Channel into 8 ohms
@ 20 HZ
Po = 0.5W, THD = 0.35%, Gain = 33.4 dB
Po = 1W, THD = 0.45%, gain = 33.4 dB
Po = 3.7W, THD = 5.0%, gain = 32.5 dB
@ 1 KHz
Po = 0.5W, THD = 0.14%, Gain = 35.1 dB
Po = 1W, THD = 0.20%, gain = 35.0 dB
Po = 4.0W, THD = 5.0%, gain = 34.2 dB
@ 20 KHz
Po = 0.5W, THD = 0.11%, Gain = 36.5 dB
Po = 1W, THD = 0.17%, gain = 36.5 dB
Po = 4.4W, THD = 5.0%, gain = 34.8 dB
This blows away the 2A3 version. Gain flatness is quite good and the output transformer behaves better at 20 Hz with the lower power output. Clipping is symmetrical and smooth when the amp is overdriven. Although listening tests were in mono only so imaging could not be really tested I was able to compare to the 2A3 left channel. The sound was more open and crisp and quite airy (typical of 45's) than the 2A3's. Serious listening tests will be conducted when I convert the right channel (I hope by tomorrow).
47's are available and new are in the $20-30 range ($40 for globes). One good thing about this tube was that it was manufactured in the tens of millions in the 1930's and 40's so there are some around. I'll have photos tomorrow and I'll send them to Brian.

then on 2/20

Posted by Lance on February 20, 2003 at 10:37:15

I finished mofifying the left channel for type 47 ST's. Results were similar to the right channel and the listening tests were equally impressive. I have sent photos to Brian so he can post them. Here is the left channel info
(into 8 ohms)
@20 Hz
Po = 0.5 W, THD = 0.45%, Gain = 34.5 dB
Po = 1.0 W, THD = 0.58%, Gain = 34.5 dB
Po = 3.5 W, THD = 5.0%, Gain = 33.6 dB
@1 KHz
Po = 0.5 W, THD = 0.35%, Gain = 36.2 dB
Po = 1.0 W, THD = 0.54%, Gain = 36.1 dB
Po = 4.0 W, THD = 5.0%, Gain = 35.5 dB
@ 20 KHz
Po = 0.5 W, THD = 0.26%, Gain = 37.4 dB
Po = 1.0 W, THD = 0.36%, Gain = 37.5 dB
Po = 4.3 W, THD = 5.0%, Gain = 36.3 dB
Gain matching between the two sides is within about 1 dB.
I changed the plate to screen resistors to 100 ohms @ 1W from 82 ohms at 1/2 W because the 100 ohm resistors look better and it is then easier to calculate the screen current (no calculator needed!)
Let's talk about the 47 for a minute. The data sheet says that the maximum plate and screen voltages should be 250 VDC. A physical examination of the tube shows that it should handle more than that since the 46 has a maximum rating of 400 VDC. I called a friend of mine who is a retired tube designer at Eimac and he knows the 47 well from his early career (1940's). He said that it should be OK at the ~280 VDC that I am running here. The 47 is very similar to the 46 but, as I said in my earlier post, but it has a different grid structure and is a pentode instead of a tetrode. The filament and plate seem identical. The 47 has a bigger plate assembly then the 45 and the 45 is rated at Pd = 10W. I am guessing that the 47 is about 11-12W Pd as this specification is not on any data sheets. Finally, the current of 39.5 MADC is divided between the plate and the screen which are tied together for low mu triode operation (I'm guessing that the Mu ~ 6). In this case screen current is about 6 MADC which makes the plate current 33.5 MADC. This results in the screen dissipation being at 1.7W and the plate dissipation at 9.5W. These should well be within the tube design centers.
KYW mentoned that when he ran the 45's in his Joplin he had to use only one 2.5 VAC winding per pair of tubes since if he kept the normal configuration that would result in too high a filament voltage. I did not see that here in the U.S. as the 47 has a slightly larger size filament (2.5 VAC @ 1.75A) than the 45 (2.5 VAC @ 1.50A). My filament voltage at Vin = 123 VAC was exactly 2.50 VAC using the original four filament windings. There could also be a difference that is due to slightly different ratios caused by differences between the taps for 220 - 240 VAC operation in the UK and 123 VAC operation here.
The 47 is also a great looking tube and as soon as I can find four new globe 47's I'll put them in. One other neat little point about the globe 47. The gettering that was used in this tube can cause it to flouresce slightly under operation as stray electrons strike the glass. This does not mean that the tube is gassy (a gassy tube would have a blue glow within the tube elements). It looks very nice if you can get lucky and have four tubes that do it.
The sound, well I'm going through listening tests now, but initial results are certainly better than the 2A3's and very similar to that which could be obtained with 45's.

Another 2A3 design by Lance - Is it what a Quad II always wanted to be?

Brian,

I thought that I would pass along some info on one of my 2A3 designs. As I said elsewhere on the list, I have done many of them over the years. The last one was as a consultant for a fellow who wanted to go into the audio biz but then pulled back when a plethora of tube amps hit the market several years ago. This design is from 1999. I bought the design back from him and have had it on the shelf since hoping to resell it when the market is better.

As you know the impetus for design sometimes really is quite obscure and this one is no exception. The main design problem that was presented to me was to make a class A monoblock with a small physical footprint. My client had obviously done his homework and he felt that there would be a market for an amp that one could fit into tight spaces. It measures only 5" x 13" without component crowding.

Before giving a brief discription about the design I would like to talk about the 2A3 a bit. This RCA design dates from 1933 and is probably one of the best audio tubes ever built. It was done as a more modern successor for the 45 and was envisoned as being run single-ended class A or push-pull class AB1. Another favorite topology was push-pull pairs. Another frequent use of this tube that most audio enthusiasts do not know about was as a speech amplifier for high power AM rado transmitters. The speech amplifier is essentially the audio input stage for the high power (up to the hundreds of kilowatts)audio modulator for the transmitter's RF power amplifier. The most accepted configuration for the speech amp was a push-pull pair of 2A3's running class A. The modulator final amps would almost always run class B. Depending on the power level there would be one or more intermediate audio satges. I mention this to make the point that there was really nothing new developed about the 2A3 after about 1940. "Design" using the 2A3 is so well documented that a few simple ohm's law calculations, if that, and your're there. It is one of the most fully characterized tubes ever made.

Modern 2A3 (last 10 years) tubes had a bit of a learning curve from a manuafcturing standpoint. Early on (1997 or so) the only choice for current manufacture was the Sino 2A3 (Chinese). These tubes were cheap and in general operated OK. There was one nagging problem, however. After getting to operating temperature a very low level "ticking" sound could be heard from some tubes. This was highly variable to the tube lot. After much conversation with my contacts in the tube industry the supposition was that the noise was thermomechanical being caused by the assembly design of the twin plate structure. I was buying them 25 at a time and returning 10-20 of them beause of this probelm. Today the learning curve is back up and we have better choices. Current Chinese tubes are better.

The design of the pictured amp (I've attached several photos) is simplicity itself. A single section of a 6SL7GT as a voltage amplifier is driving the 2A3. Coupling is capacitve. This simple topology was driven by the fact that I didn't have a great deal of room to work with. Don't pooh-pooh it though. The 6SL7 is a beautiful sounding tube with lots of voltage gain making it perfect for this application. This design, by its nature, is very critical as far as the coupling cap is concerned. You can change the sound of the amp markedly by changing this capacitor. The one shown in the photo is a Solen and, at least to my ears, sounds the best (it's also the cheapest). I have also tried many of the other more expensive caps and all work fine but with differing coloration. The rectifier shown is a 5V4G but the commercial implementation of the design was to use a Golden Dragon GZ37 (it also looks nice). I like this tube as it has a long warmup characteristic that tracks the 2A3 warmup cycle pretty well. For my own tastes, I find that rectfiers that incorporate a distinct cathode sound better. The output transformer is an Atwood. The bottom of the chassis contains the filter choke and 2A3 filament transformer. One rule in this design, it must use currently available and manufactured parts.

Another planned characteristic of this amp was that a 300B could be substituted during manufacture with no effort. Since the filament transformer is a separate item all that was needed was to swap it out with a 5V unit during assembly. Now, of course, the 300 B would be running at about 250V (same as the 2A3) and only put out about 4W. I like this tube much better when run a lower voltages then the incandescent point that most of them are run at today. They will also last much longer...no small item at their price. The hum level is very slightly higher but that is usually normal when comparing the 2A3 versus the 300B. In the 2A3 version this amp has a very open airy sound and is very pleaseant to listen to. The 300B variation has a richer more complex sound but is not quite so open. I have included a photo showing the amp with a Sovtek 300B also. The production rated power is 3.5W. There is no global feedback.

What you see in the photos is the final preproduction version. This was done on a commercially available chassis but the production version was to be punched and then bent. The two jacks at the side of the amp are for bias point checking. The two pots are for bias adjustment and hum balance. The fianl production version was to be chromed along with the transformer bells.

Brian, please feel free to post this if you want.

Regards,
Lance