Scratch build valve power amps

The amp has been working great, doing 10-14 hours a day almost every day for nearly 4 months now.


I finally managed to get a decent picture showing all of the grids with decent depth of field.


The compacted MkIV version is still in progress. Time and the weather have not been on my side. I've done 90% of the wiring and I've got the front and back panels ready to fit. I just need to mount them and then wire up the mains input and switch. I also need to wire the input socket via the high pass filter board which will also need to be mounted. It sounds like a lot but it shouldn't take more than an afternoon to sort.
 
I've finished fitting the front and back along with the divider. The capacitors are all in place too. Not the prettiest as the wood is reused from previous builds. (the black on the back panel is where it got used to support something being painted)


Most of the wiring is now done. Hopefully I can have this finished tomorrow so I can test it.
 
Last edited:
Not tempted to 'ultra linear' the long tail pair you have there? (http://www.tubecad.com/january2000/page11.html) .... #nowresisttheurge :p (actually I'm not entirely sure what they're doing at that point.. I think it may be witchcraft)
I think that is a bit too complicated for my experience level and the additional valves, whilst pretty, only serve to use more power and produce a lot more heat. It would be nice to achieve it using a hybrid with solid state parts.

Interesting you have issues with the 6SN7 - the TungSol datasheet states that the cathode should be attached to the heater and that the voltage difference between the heater and cathode should be kept as low as possible. Sheet here:https://frank.pocnet.net/sheets/127/6/6SN7GT.pdf

Interesting as I've never heard that about other SN7s.
Not sure I've had any issues with 6SN7, though I use the equivalent noval 6CG7/6FQ7. The only valve I've had problems with is the input pentode. One EF806 and a single 6059 both of which produced large amounts of hiss.

I finished the Mk IV prototype and have now tested it. I made the usual mistake of reversing the phases leading to positive feedback, but I've managed to do it 3 times now so know how to quickly fix it. Takes 2 mins with a soldering iron to reverse the outputs from the phase splitter. Tested on the bench, the performance appears to be very similar to the other amp which is excellent.

1KHz square


5KHz square


10KHz square


It's a good looking amp, which would be even better with some decent wood panelling. (But I'm not wasting a good piece of oak or similar on a prototype)


Working together with the Mk III. First time having both speakers driven by my valve amps. Doesn't sound half bad. I've currently not got anywhere to put the second amp so it had to go on top of the speaker. (the Kef isn't in use)
 
Bought some higher voltage rated capacitors with some matching capacitor mounts and more octal sockets. The F&T 500V rated capacitors may be reused to replace one of the 10uF capacitors. The additional belton octal sockets will replace the old finder sockets I've used on the rectifier as the pins are a bit loose.
 
In the dim and distant past (in my Student youth) I have built "HiFi" amplifiers with a few transistors from designs in "Practical Wireless". These could probably be called "simple solid state amplifiers" but I doubt if they were really hi fidelity. Now, if I look inside my Sony "scala" hifi amplifier (bought a fee years ago) it looks highly complex, with ICs as well as power transistors. Presumably (?) this complexity is there for some reason??

I do not know, but the "best" (i.e. the most expensive) kit, which is meant to provide the purest sound replication (?) does not look like "simple solid state" gear to me. I am not doubting your desire, just wondering whether it is achievable in practice?
Before I moved into the realm of vacuum tubes, I had done some heavy leg work on building a Quad 606 clone. As amps go, for the power output, the quad current dumping design is relatively simple in terms of size and component quantity. The Naim NAP250 is commonly cloned and that seems to be a very simple amp with a very good rep.

I finished off doing the second one this morning.


Doesn't look too bad considering the cobbled together nature of these prototype builds. Once I've figured out how I'm going to get the proper aluminium chassis made, I can get them built properly.


I've also invested in some DC blocking PCB's and all the necessary components to build them. I've always had some residual hum on all five of my amplifiers which has made me suspicious of there being some slight DC offset.


I've made a start on building one of them. I got 3 PCB's so I can do other parts of the house.
 
Thanks, I finished fitting the diodes this evening. Just the box wiring left to do now.


Those large tracks absorb a heck of a lot of heat and are take a while to solder even with a 70W iron.
 
I see diodes and caps, essentially a smoothing network. The issue I also see is there's no bleed resistor for the caps, so when you switch off it holds charge.
Doubt it matters in this case as they are only 16V capacitors. My valve amps don't have bleeders either. When switched off, the caps discharge completely. I built a Carlson's Lab style capacitor discharge tool to be sure when dealing with capacitors anyway.

I can report that it has made the transformers quieter on all of the amps which is nice.
 
N


Cool -- I just had visions of 240V (due to the earth wire) caps charged. With the caps charged do you not find you have HT over the valves without the heaters/grid operational?

I've sort of pretty much decided on a simple single end design initially for mine, but I'm still exploring.
This is the schematic for the DC trap. https://sjostromaudio.com/hifi_pics/hifi_100pr/dct03r0schema_p1color.png

The caps in the valve amps are so small (47uF is the biggest, the rest are just 10uF) that they barely last a full second on shut down, so the HT collapses very quickly.
 
QQ did you opt for a regulator in your DC HT power supply?

I’m considering if I should put something after the smoothing caps but before the HT power distributions (it will have filter caps for each tube or stage)
There is no regulation in my amp. B+ sags by about 20V once there is a decent level output. Maybe a future upgrade along with dc heaters for the first stages.
 
Added another pair of JJ KT77's and JJ GZ34 to complete the collection. I also picked up a pair of 6BR7 as they are no longer made so having a decent stash is going to save having to rewire the socket back to an EF86 for a few years. (CV4006/6059 is just the industrial version of 6BR7)


The collection has gotten large enough that I'm now considering a plastic storage box for them.
 
Turns out, one of the 6BR7 was faulty. It produced virtually no gain making one amp very quiet vs the other. After some actual measurements today, I can see why.

Static Voltages with 0V at G1:

Faulty 6BR7
Anode: 236V
Cathode/G3: 0.95V
G2: 236V

Other supplied 6BR7
Anode: 105V
Cathode/G3: 1.98V
G2: 96V

Known good 6BR7 from my own collection
Anode: 88V
Cathode/G3: 2.1V
G2: 86V

Oscillograms with 150mV 1KHz sine input: CH1 is speaker out, CH2 is anode out of the 6BR7. Explains why one channel was extremely quiet as the output voltage was almost a quarter of what it should have been. In power terms, 2.86V into 8 ohms from a good valve is a shade over 1W. 0.76V into 8 ohms is a measly 0.07W


Suffice to say, it's on it's way back to the seller.
 
After someone mentioned trying to build a Mullard 5-20 using 807 output valves on DIYaudio, I got it into my head to look at whether it was possible using my spice simulation as the ideal output transformer for 6L6 based valves is 6.6K. (which is what the 5-20 happens to use) Turns out, based on my simulation, 6L6GC and it's 30W+ derivatives will work just fine in my amp. So on a whim, I bought some of the high dissipation (35W) 7851A's to try out.

With all of my output valves side by side, they look pretty similar to the EH 6CA7 on the far left, but the primary difference is that the cathode is quite a bit smaller. It does seem odd that it has a 35W dissipation limit whilst the rest are all 25W.


Fitted and running. Cathode biasing seems to work nicely. Plates sit at 428V and screens just below at 427V via the 43% ultra linear taps. Small bonus these have is the reduced heat output from the lower heater current (0.9A vs 1.5A) After a couple of hours listening, it sounds every bit as good as the other EL34's and their derivatives. I figure the ~20dB of negative feedback is minimizing any audible differences there may be between the various valves I've tried. A big added aesthetic bonus is that these have the blue glow on steroids compared to the rest.
 
Last edited:
Quick question - which solder did you go for?

I’ll be using ptfe 1000V copper - I can’t remember if that was silver coated copper but if it was I’m going to have to go with a couple of % silver mix - Mundorf do an on free mix with 3.8%.
Sorry for the slow reply, I didn't get a notification of your message. I used standard multicore solder 60/40. I have always stuck with lead solder as the joints seem to last exceptionally well. I've not had any trouble with the amps so far and I doubt I ever will. It would have been lead solder back in the day so I don't see an issue.
 
Back
Top Bottom