I find the Heath AA-32 a pleasant little amplifier perfect for a office or den.
It is small and simple, gives off moderate heat and in my experience they are long lived and reliable.
Although simple and modest in design and measured performance I find the AA-32 produces a utterly pleasant musical experience when called on to fill in the acoustic spaces in ones den or office with pleasant background sounds to work by. While the output quad of NOS 6GW8 tubes with many hours on them still produces a acceptable amount of output power at moderate distortion they have now lost about half their emissions. One can see the writing on the wall for this set of tubes. So what to do as for me the prices of a set of NOS 6GW8 tubes just does not compute.
My first tact was to switch the outputs with the available and low cost NOS PCL86, the same tube with 13V heaters. Searching for a available small and affordable filament transformer that would fit under the chassis to provide the 13V for the 4 PCL86 proved harder than I expected. There is little space under the AA-32 chassis. I did find a 7V 10VA (6.3V+7V=13.3V) unit from Amgis part number XF-10224 that is 23.5mm thick and should just fit under the chassis however there seems the be a minimum order of 10 pieces for a total cost of over $250 and a 26 week lead time. If anyone knows of a source of this transformer or a replacement that can be ordered in singles let me know as this is still a possible way to go.
As part of my preparation for changing the output tubes I figured to audit the passive parts in the output stage. The result was there were almost no resistors and few capacitors that were still within the original specification. Resistor value drift of 30% was all over the amplifier's power section and it is testament to a robust and forgiving design that the amplifier still sounded and measured close to the specifications of the unit when new. It did seem clear that any tube replacement would need to involve replacement of almost all the power sections resistors and capacitors. Given all the parts had to be removed and replaced I figured 4 new sockets were in order. Can you see where this is going?
Now looking at rebuilding the output section from scratch why stick to 6GW8 tubes? How about replacing the original design with a new design using other tubes. Yes it would no longer be original a valid concern but nether will it be original with a extra power transformer stuffed under the chaise.
Looking around what I had on hand I came up with a bunch of Russian 6F5P, a power pentode and low level triode combo designed for the TV industry as a vertical sweep amplifier.
These have a max plate and screen voltage of 250V close enough to the 290 and 280 the AA-32 runs at. When considering sweep tubes, ratings can often be pushed in audio amplifier service in my experience. The 6GW8 data sheet lists S=10.5 and the 6F5P data sheet has S=7 suggesting the 6F5P would lack some gain, a possible issue as the simple drive circuits in the AA-32 are pretty limited in both voltage swing and gain.
Peak plate current @ a=75V g1=0V is listed on the data sheet curves for the 6GW8 at about 107mA.
Looking as the 6F5P peak current is shown as.. well plate current is not show as it is off the charts high @ over 200mA and this is with only g2=170V? What would happen with g2=250V or more? The 6F5P it seems is a sweep tube with a very sensitive screen not at all like the 6GW8 and clearly dropping it into a AA-32 like circuit will not go well. Add in the triode in the 6F5P has a u= 70 VS the 6GW8 triode u=100. With lower gain and limited drive into a pentode needing higher bias voltage and more drive the 6F5P seemed doomed.
Having owned a utracer for some years I have leaned the data sheet tells only a very limited story about a tube. There are many more operation conditions for a tube than what is listed on the data sheet that can be of interest if you do not mind straying off the beaten path.
So I ran some curves on the 6F5P and the 6GW8 looking at peak plate current VS screen voltage and found that the 6F5P hit the same g1=0V peak plate current at g2=100V as the 6GW8 hit at g2=280V. So the 6F5P can deliver the same peak plate current as the 6GW8 with only 100V of screen voltage. Hmmm.
Next I ran the 6F5P plate current VS the g1 voltage with a=300V g2=100V.
The 6F5P result for a=30mA was a g1=-9V the exact same g1=-9V for a=30mA as a 6GW8 with a screen voltage of about 280V.
With the same peak plate current and the same bias voltages suggests a 6F5P with a screen voltage of 100V has almost the same characteristics as a 6GW8 at a screen of about 280V.
I build up a spice simulation of the current AA-32 circuit and a new version using the 6F5P with a screen of 100V both using a 290V plate voltage. The result was both versions driven with the same drive voltage delivered the same power output into a spice model of the AA-32 output transformer with about the same THD. This was fun to see.
Now lets see of we can replace the driver stage with 6F5P triodes u=70 VS the u=100 of the 6GW8. I was hopeful as the 6F5P triode has S= 7 to a low S=1.6 of the 6GW8 and triode realizable gain is influenced by S as well as u. The AA-32 input stage gain simulated out at 33.208dB @2.14% THD and my version with the 6F5P triode resulted in a gain of 34.717dB @0.75% THD with no feedback. Looking good.
As the 6F5P triode has high S there was no need to bias it separately like the original AA-32 for the split load phase splitter so I lost the blocking capacitor and grid leak resistor between the gain stage and the splitter of the AA-32 design saving a few parts, improving the low frequency phase margin and dropping THD.
So the 6F5P version came out with a gain 1dB less and lower in THD. Very workable with a caveat. The 6F5P due to sensitivity of screen voltage changes would need some sort of screen voltage stabilization circuit in operation to set the screen voltage and so the operating point.
Below I have attached the spice circuit diagrams for both the original AA-32 and the 6F5P version.
For me the jury is out on how I will deal with my aging AA-32 output tubes.
The above may be useful for others that have a unit they would like to put back into service that uses the 6GW8 or anyone who wants a nice simple good performing amplifier based on the 6F5P.
It is small and simple, gives off moderate heat and in my experience they are long lived and reliable.
Although simple and modest in design and measured performance I find the AA-32 produces a utterly pleasant musical experience when called on to fill in the acoustic spaces in ones den or office with pleasant background sounds to work by. While the output quad of NOS 6GW8 tubes with many hours on them still produces a acceptable amount of output power at moderate distortion they have now lost about half their emissions. One can see the writing on the wall for this set of tubes. So what to do as for me the prices of a set of NOS 6GW8 tubes just does not compute.
My first tact was to switch the outputs with the available and low cost NOS PCL86, the same tube with 13V heaters. Searching for a available small and affordable filament transformer that would fit under the chassis to provide the 13V for the 4 PCL86 proved harder than I expected. There is little space under the AA-32 chassis. I did find a 7V 10VA (6.3V+7V=13.3V) unit from Amgis part number XF-10224 that is 23.5mm thick and should just fit under the chassis however there seems the be a minimum order of 10 pieces for a total cost of over $250 and a 26 week lead time. If anyone knows of a source of this transformer or a replacement that can be ordered in singles let me know as this is still a possible way to go.
As part of my preparation for changing the output tubes I figured to audit the passive parts in the output stage. The result was there were almost no resistors and few capacitors that were still within the original specification. Resistor value drift of 30% was all over the amplifier's power section and it is testament to a robust and forgiving design that the amplifier still sounded and measured close to the specifications of the unit when new. It did seem clear that any tube replacement would need to involve replacement of almost all the power sections resistors and capacitors. Given all the parts had to be removed and replaced I figured 4 new sockets were in order. Can you see where this is going?
Now looking at rebuilding the output section from scratch why stick to 6GW8 tubes? How about replacing the original design with a new design using other tubes. Yes it would no longer be original a valid concern but nether will it be original with a extra power transformer stuffed under the chaise.
Looking around what I had on hand I came up with a bunch of Russian 6F5P, a power pentode and low level triode combo designed for the TV industry as a vertical sweep amplifier.
These have a max plate and screen voltage of 250V close enough to the 290 and 280 the AA-32 runs at. When considering sweep tubes, ratings can often be pushed in audio amplifier service in my experience. The 6GW8 data sheet lists S=10.5 and the 6F5P data sheet has S=7 suggesting the 6F5P would lack some gain, a possible issue as the simple drive circuits in the AA-32 are pretty limited in both voltage swing and gain.
Peak plate current @ a=75V g1=0V is listed on the data sheet curves for the 6GW8 at about 107mA.
Looking as the 6F5P peak current is shown as.. well plate current is not show as it is off the charts high @ over 200mA and this is with only g2=170V? What would happen with g2=250V or more? The 6F5P it seems is a sweep tube with a very sensitive screen not at all like the 6GW8 and clearly dropping it into a AA-32 like circuit will not go well. Add in the triode in the 6F5P has a u= 70 VS the 6GW8 triode u=100. With lower gain and limited drive into a pentode needing higher bias voltage and more drive the 6F5P seemed doomed.
Having owned a utracer for some years I have leaned the data sheet tells only a very limited story about a tube. There are many more operation conditions for a tube than what is listed on the data sheet that can be of interest if you do not mind straying off the beaten path.
So I ran some curves on the 6F5P and the 6GW8 looking at peak plate current VS screen voltage and found that the 6F5P hit the same g1=0V peak plate current at g2=100V as the 6GW8 hit at g2=280V. So the 6F5P can deliver the same peak plate current as the 6GW8 with only 100V of screen voltage. Hmmm.
Next I ran the 6F5P plate current VS the g1 voltage with a=300V g2=100V.
The 6F5P result for a=30mA was a g1=-9V the exact same g1=-9V for a=30mA as a 6GW8 with a screen voltage of about 280V.
With the same peak plate current and the same bias voltages suggests a 6F5P with a screen voltage of 100V has almost the same characteristics as a 6GW8 at a screen of about 280V.
I build up a spice simulation of the current AA-32 circuit and a new version using the 6F5P with a screen of 100V both using a 290V plate voltage. The result was both versions driven with the same drive voltage delivered the same power output into a spice model of the AA-32 output transformer with about the same THD. This was fun to see.
Now lets see of we can replace the driver stage with 6F5P triodes u=70 VS the u=100 of the 6GW8. I was hopeful as the 6F5P triode has S= 7 to a low S=1.6 of the 6GW8 and triode realizable gain is influenced by S as well as u. The AA-32 input stage gain simulated out at 33.208dB @2.14% THD and my version with the 6F5P triode resulted in a gain of 34.717dB @0.75% THD with no feedback. Looking good.
As the 6F5P triode has high S there was no need to bias it separately like the original AA-32 for the split load phase splitter so I lost the blocking capacitor and grid leak resistor between the gain stage and the splitter of the AA-32 design saving a few parts, improving the low frequency phase margin and dropping THD.
Code:
Below is a chart of the gains achieved.
Stage AA-32 original 6F5P redesign.
Input 33.208dB 2.14% THD No FB 34.717dB 0.75% THD No FB
Phase splitter top 32.828 3% 34.39 1%
Phase splitter bot 32.838 34.39
Output plate 64.998 64.03
Output 8 ohm Spk 39.432dB 3.18% THD No FB 38.464dB 2.25% THD No FBSo the 6F5P version came out with a gain 1dB less and lower in THD. Very workable with a caveat. The 6F5P due to sensitivity of screen voltage changes would need some sort of screen voltage stabilization circuit in operation to set the screen voltage and so the operating point.
Below I have attached the spice circuit diagrams for both the original AA-32 and the 6F5P version.
For me the jury is out on how I will deal with my aging AA-32 output tubes.
The above may be useful for others that have a unit they would like to put back into service that uses the 6GW8 or anyone who wants a nice simple good performing amplifier based on the 6F5P.
Last edited:
One thing to note is that the PCL86 will run perfectly well at 12.6V. There should be plenty of transformers available, or even a small smps.
PS You need only 6.3V extra, you already have 6.3V from the original transformer, right?
PS You need only 6.3V extra, you already have 6.3V from the original transformer, right?
Last edited:
Good point on the 12.6V transformer.
With a 12.6V @1.2A = 15.2VA the size is larger than the smaller 7V @1.2A = 8.4VA or 10VA unit.
Importantly I could not find a 15VA unit that was thinner than 25mm (the inside chassis depth)
The 10VA unit is 23.5mm deep so just fits inside.
You are right here are lots of 12.6V transformers available if I a willing to it put the transformer above the chassis but considering the transformer size and the small space up top I considered it seemed a ugly fit on top of the chassis.
Perhaps I need to revisit this and be less picky.
There are a number of smaller 12V DC switch supplies on ebay but both reliability and high frequency noise was a concern. Pretty sure any high frequency hash on the DC in the filaments will end up in the audio path.
I personally think this lack of high frequency noise artifacts in older tube stuff is one reason they do not induce "listener fatigue" with long hours of use.
More thinking needed here it seems as I do like the idea of staying with a more original circuit if possible.
With a 12.6V @1.2A = 15.2VA the size is larger than the smaller 7V @1.2A = 8.4VA or 10VA unit.
Importantly I could not find a 15VA unit that was thinner than 25mm (the inside chassis depth)
The 10VA unit is 23.5mm deep so just fits inside.
You are right here are lots of 12.6V transformers available if I a willing to it put the transformer above the chassis but considering the transformer size and the small space up top I considered it seemed a ugly fit on top of the chassis.
Perhaps I need to revisit this and be less picky.
There are a number of smaller 12V DC switch supplies on ebay but both reliability and high frequency noise was a concern. Pretty sure any high frequency hash on the DC in the filaments will end up in the audio path.
I personally think this lack of high frequency noise artifacts in older tube stuff is one reason they do not induce "listener fatigue" with long hours of use.
More thinking needed here it seems as I do like the idea of staying with a more original circuit if possible.
Last edited:
Just for me to understand. You already have the amp, right? If that is true, then you only need extra 6.3V, connected in series with the existing 6.3V winding. And only half the power, the other half is covered by the existing transformer.
Re:"Just for me to understand. You already have the amp, right?"
Correct I have a old and trusty AA-32 in my collection that is needing the output tubes refreshed and I am looking at ways to get there without handing over my life saving for a quad of NOS 6GW8s. I am looking for a solution that I can use hours daily for may more years to come. So all the current tired and out of spec. output components will all go and then the question become what to replace the output circuits with?
Original, a bit different or a clean slate.
Correct I have a old and trusty AA-32 in my collection that is needing the output tubes refreshed and I am looking at ways to get there without handing over my life saving for a quad of NOS 6GW8s. I am looking for a solution that I can use hours daily for may more years to come. So all the current tired and out of spec. output components will all go and then the question become what to replace the output circuits with?
Original, a bit different or a clean slate.
Last edited:
Then you just need extra 6.3v 1.5A, to be safe. Combine that with the 6.3V of the existing transformer.
Availability of one that is small enough to fit inside is the challenge.
There is not a lot of room inside the AA-32 and 10VA units I have not found a source for yet in either 6.3V or 7V units.
To fit under the chassis it needs to be less that 24mm high as noted in my first post with the Amgis part number XF-10224 that is 23.5mm thick.
There is not a lot of room inside the AA-32 and 10VA units I have not found a source for yet in either 6.3V or 7V units.
To fit under the chassis it needs to be less that 24mm high as noted in my first post with the Amgis part number XF-10224 that is 23.5mm thick.
Isn't that just a 'feature' of insufficient filtering? Most modern high end test equipment (oscilloscopes, frequency analysers and generators e.t.c) use switch mode power supplies. If it's good enough for Siglent, shouldn't it be good enough for a Hi-Fi? Capacitors and inductors might be easier to source/accommodate...
And you’re going to buy one of that quality, on EBay, for less than a the cost of four NOS 6GW8’s, right?
Simply adding LC filtering (and Ls are not cheap) to a cheapie wont fix the design problems that result in all that excess hash either. It will find its way around that stupid filter.
Simply adding LC filtering (and Ls are not cheap) to a cheapie wont fix the design problems that result in all that excess hash either. It will find its way around that stupid filter.
Could you not knock up a voltage doubler. That'll give you 12.6v at just under half the current which is 660 ma for a 6GW8. You need 300ma for a PCL86.
You could switch it too so it will accept both voltage tubes.
Also using EL86 would be a lot more inexpensive, Watford valves do new Tunsgsram for £35 a piece, no idea of postage cost or if they ship to Canada though.
You could switch it too so it will accept both voltage tubes.
Also using EL86 would be a lot more inexpensive, Watford valves do new Tunsgsram for £35 a piece, no idea of postage cost or if they ship to Canada though.
Last edited:
Any reason not to put a barrel socket on the back and use a wall adapter for the 14GW8/PCL86 heaters?
(There might be one but I don't know what it would be.)
Meanwhile keep watch for a power transformer with all the right voltages and form factor to swap in place of the original.
(There might be one but I don't know what it would be.)
Meanwhile keep watch for a power transformer with all the right voltages and form factor to swap in place of the original.
Crazy idea - that would work and something I would be inclined to try…..
You need a very SMALL transformer? You need 16 miserable watts of heater power and no room to put a 50 VA unit, right? If you can GET a transformer core just big enough for the 16 watts, and make a little autotransformer to go from 6.3 volts to the required 13.3. Wont need 50 billion turns or even any isolation. Even if you needed 10 or 15 turns per volt to get it mechanically quiet it should be easy to make, and you wont go to sleep counting turns. All that overhead associated with line voltage can be dispensed with, making it pretty damn small.
You need a very SMALL transformer? You need 16 miserable watts of heater power and no room to put a 50 VA unit, right? If you can GET a transformer core just big enough for the 16 watts, and make a little autotransformer to go from 6.3 volts to the required 13.3. Wont need 50 billion turns or even any isolation. Even if you needed 10 or 15 turns per volt to get it mechanically quiet it should be easy to make, and you wont go to sleep counting turns. All that overhead associated with line voltage can be dispensed with, making it pretty damn small.
I don't see any space for a transformer above or below chassis - the 6F5 (US 6GV8) may be your best bet (and 6BM8s aren't as hard to find; another option). Screen voltage can be stabilized with a few 5W zeners in series. Voltage doubler would add quite a bit to the load on the heater winding, voltage doubles but current is about triple.
Could you add some more turns of copper wire around the existing power transformer? How does it look? Is there space between the bobbin and the laminations?
Otherwise I see surplus toroids for 15-20VA pop up on my local auction site. Normally they are quite slim at that power level. Also there are tiny booster power supplies that could be driven from low voltages, but then efficiency might be a constraint.
Otherwise I see surplus toroids for 15-20VA pop up on my local auction site. Normally they are quite slim at that power level. Also there are tiny booster power supplies that could be driven from low voltages, but then efficiency might be a constraint.
I have a stash of NOS Tungsram Australia ECL86´s. Can sell 8 of them at a reasonable price, not as cheap as PCL´s though. Can select in my AVO. Message me if interested.
Have you looked at Antek for transformers?
https://www.antekinc.com/10va/
Not sure what they will cost to ship, but I feel like it wont be too bad. Then you can simply run the different filament voltage alternatives.
https://www.antekinc.com/10va/
Not sure what they will cost to ship, but I feel like it wont be too bad. Then you can simply run the different filament voltage alternatives.
Wow, the is a big upgrade. I have one that is still all original from the time a GM engineer built it when it first came out. When we moved 7 years ago, I dropped it and bent the chassis and all the tubes broke.