First valve amplifier build

[cathoderay57]

I can't agree with Post #39. A.M. Ball's Radio Valve and Transistor Data lists output valves operating in push-pull. While the data column's title for Rk states (per valve (Ohms)) there is a postscript against the actual value that says "common resistor" which in the case of 6K6 is 400 Ohms. Each valve requires the same grid bias voltage and for a common cathode resistor it will carry 2 valves' worth of current. Therefore for individual cathode biassing for the same grid bias at half the current then twice the resistance is required i.e 800 Ohms.

I agree lower value resistors can be used for the 6K6 grid1 couplers but I suspect it won't make any noticeable increase in sound output.

The output transformer could possibly have been intended for a 3 Ohm speaker and since you are using an 8 Ohm unit you are possibly losing a bit of output power as a result. Whether you can find a decent 3 Ohm speaker with the required power handling at a reasonable price is another issue of course. When you add negative feedback you'll lose a bit more but you still have the other half of the ECC83 to play with.

Cheers, Jerry

[ms660]

Going by the valve data in Post#39 I make it 400 Ohms individual in that valve data and 200 Ohms for shared.

Lawrence.

[snowman_al]

Easy way to be sure is measure the voltage drop across the cathode resistors as I mentioned in post #37. You can work out the current draw then and see if it is in 'spec' or not.

[cathoderay57]

Well, I won't die in a ditch over it. Ball's data attached. I guess the acid test is fire it up and measure the cathode voltages then compare currents to the data. Jerry

[ms660]

Yes, my mistake, sorry about that, just to add that in the valve data book I have it shows the same postscript for the Brimar but doesn't show any for the Ferranti.

Lawrence.

[snowman_al]

And just to add to the confusion here is a BRIMAR 6K6 entry without the 'foot note'. Iliffe books as well.

There are clearly some errors and or inconsistencies in the BRIMAR entries, for instance on Jerry's pdf the EL41 is also shown as 140Ω with the shared resistor note, but individual 140Ω resistors in the Mullard data sets... + others.

So anyone's guess for the 6K6.
Alan

PS, sorry Gabe it must be totally confusing.

[Gabe001]

So, the 47k resistor to ground suggested by Alan really helps, but there's still a tiny bit of hiss so I need to experiment in that area, but heading in the right direction.

6k6 grid stopper changed from 220k to 10k - not much difference. Possibly a bit louder but may be recall bias.

The voltages and current based on the readings from 6k6s with the 870R resistors are: (measured with DMM)

Anode 266v. Voltage drop across OPT 10v. DC Resistance of OPT 400R = anode current 25mA

Screen 260v.

Cathode 21v ie 21v across 870R = 25mA cathode current

I believe that this is according to spec?. Readings taken at max volume but no input signal.

[cathoderay57]

Hi Gabriel, 25mA cathode current (per valve) is a little bit lower than I'd have expected when you sum the expected anode and screen currents. Are both valves measuring the same, and what screen current are you measuring? I presume you meant an 820 Ohm cathode resistor, not 870 Ohm? Cheers, Jerry

[Gabe001]

Yes I did mean 820 apologies.
2 valves measure exactly the same.
Based upon the voltage drop across the first 1k resistor, the combined screen current for both valves is 16mA

[cathoderay57]

Well the first 1k resistor also carries anode current for the 6C4 and ECC83 (both totalling approx 5mA) leaving about 5mA screen current per 6K6. I'd say that is about right for 6K6 screen current but it means anode current is only about 20mA which is low. Try swapping out the 820 Ohm cathode resistors for 680 Ohm and if you are still not getting a cathode current of about 30mA per valve go down to 560 Ohm. Cheers, Jerry

[Gabe001]

Thanks jerry. It's hard to calculate the screen current as there is no measureable voltage drop across the 47R feeding into the screen.

Meanwhile yesterday night I did some measurements using a dummy load. Using a cheap Bluetooth receiver as a signal source and the frequency generator app on my phone, peak to peak voltage using 4ohm resistor is 11v, which I think currently equates to a power output of almost 4w RMS at max phone and almost max amp volume

Frequency response stable between around 250hz and 17khz (very approximately) and then drops gradually
Square wave doesn't look to bad at 400Hz.

I'll try lowering the cathode resistor but due to this ambiguity related to the 6k6 cathode resistor I'm increasingly tempted to use 470R and swap to 6v6s

[Gabe001]

So I've sorted the hiss. It was the ecc83. I swapped it with another from my spares and everything is quiet now.

I'm really satisfied with how this turned out. Volume is uncomfortably loud when cranked up fully, so it's more than enough for me, even with just half of the ecc83. It sounds really nice actually!

Jerry, I don't have 680/560 ohm resistors rated 3 or 5 w
So I strapped in a 3.3k resistor in parallel with the 870ohm resistor. Its only 1w, so it is very temporary, but it's good enough for a test. This equates to a 656 ohms cathode resistor.

The readings with the 656 cathode resistor are as follows:
Readings at full volume with a signal
Anode 261v
Screen 258v
Cathode 19v
Cathode current 29mA

I think power is slightly lower than with the 820 ohm resistor, but then again I may be dreaming.

Any ideas regarding a potential resistor-capacitor combination for feedback, and what cathode resistor shall I stick with?

Thanks
Gabriel

[cathoderay57]

Hi Gabriel, well according to Ball's data the cathode current per valve for a 6K6 in push-pull can range safely between 32-36mA (the sum of anode and screen currents). If you are currently running about 29mA I would try raising a bit more so that you are within the published range which is likely to be determined as the best compromise between output power and distortion. I'll see if I can find a couple of suitable 560R resistors and stick them in the post. If the anode current then falls within range and sounds good I'd leave it at that. For negative feedback I'm not sure exactly how to calculate the component values. I'd start using the Mullard 5-10 values and see how it sounds. Circuit attached. Be ready to turn off quickly in case you've inadvertantly cross-connected the phase of the speaker output. Cheers, Jerry

[Gabe001]

Thanks Jerry. I've dropped you a PM. I'll do some experimentation with the negative feedback loop.

I've got a rather embarrassing rookie amplifier question, as I'm tempted to play with the other half of the ecc83

The question is whether adding further gain to my amp will be of any benefit since I am using only a 'high' level input that's a Bluetooth receiver. I can already drive the output pair to clipping at max volume. As such, would there be any further benefit of adding in the remaining half of the ecc83, in terms of output wattage, or is the push pull pair already at max and further gain won't translate into more output watts?

Thanks
Gabriel

[cathoderay57]

If you're getting clipping then you won't get any more power without installing a replacement mains transformer with higher HT. The 6K6 is a robust valve, also being used for linescan, but this approach would probably be a law of diminishing returns. It might still be worth trying a 3 Ohm speaker if you can find a beefy one. Otherwise go with 6V6 if you are prepared to risk the output transformer, which is unknown provenance. Adding the extra ECC83 stage is probably only worth it to compensate for a passive tone control network or to use a low output source such as a magnetic turntable cartridge. Cheers, Jerry

[Gabe001]

Thanks Jerry

I did try with 2x4ohm speakers in parallel (it's all I've got) and I've got ample volume. The problem with using the phone is that not all apps output identical volume signals, Spotify being on the lower volume end of the spectrum. In any case, I'll be fitting a krc-86b into this, which is generally louder than my phone so I think, even accounting for the NFB attenuation and the effect of reducing the grid bias further, I'll be ok. If not I'll switch to the 6v6s and take my chances.

It's been a good experience overall, and I found this whole exercise extremely stimulating. Many thanks to all those who contributed.

I'll post a video when all is done, but I've still got some work to do - NFB, base (wooden), speaker terminals, pilot lamp, fuse etc so it may be a few more weeks.

[Herald1360]

6V6's shouldn't cane things too much- 0.1A more heater load but about the same cathode current, thanks to lower g2 current, per pair.

[Gabe001]

In case this is of interest to anyone, this is the best performance I managed to get out of this amp.

HT after surge resistor 273v

I've played with both the 6k6 pair and 6v6 and finally settled on the 6v6

The RMS voltage using the 6k6 pair with a 4ohm resistor was 3.5v which equates to 3watts

The RMS voltage with the 6v6 pair with a 4 ohm resistor is 4.3 which equates to 4.6watts. This is with my phone/bluetooth receiver as a signal source. There is a bit of clipping on the oscilloscope past 90% volume.

The output transformer by my calculations has an impedance of 13.5k ohms primary and 3 ohms secondary (turns ratio 67, based upon 7.4v secondary translating to 500v primary). Its a bit high for the 6v6s which is probably why I'm not getting the best out of them. Its however loud enough as it is and I can't heat any distortion except at max volume

Hum is minimal. Hiss is gone. I'm using screened cable in the first section now, but the main resolution to the hiss came from a new ecc83.

I tried implementing NFB using the Mullard circuit as a guide. Using the same size resistors dropped the RMS voltage output by about 0.5v. Eventually I settled on a 0.1v RMS drop which required a 68k resistor. However I couldn't hear any tangible benefits apart from a bit of attenuation, so I ended up removing it.

Voltage around 6v6s as follows (full signal and full volume)
Anode 262
Screen 257
Cathode 15
Cathode resistor 470 ohms
Cathode current per valve 32mA (a bit low)

I still need to do some cosmetic work and fit a fuse but I'm going to call it a day with regards to the electronic parts. I'm happy with the outcome.

Video below, at about 50% power when it already starts to get uncomfortable. I'm using two rather rubbish 4 ohm speakers in parallel which is all I've got.

It's been a great learning experience. Thanks for the help ☺️

https://youtu.be/SWrFcE9FgBc

Gabriel

[Radio Wrangler]

Originally Posted by Gabe001

I tried implementing NFB using the Mullard circuit as a guide. Using the same size resistors dropped the RMS voltage output by about 0.5v. Eventually I settled on a 0.1v RMS drop which required a 68k resistor. However I couldn't hear any tangible benefits apart from a bit of attenuation, so I ended up removing it.

You weren't applying a tangible amount of negative feedback.

Roughly the benefit is proportional to the amount of gain reduction. Halving the gain will roughly halve your distortion, and halve your non-flatness.

In the extreme, Leak applied about 20dB of negative feedback, which was right on the limits of what could be made stable. This reduced a 1% distortion amplifier to 0.1% but it also reduced the gain tenfold from what it would do without feedback.

Consequently, you don't just 'apply' feedback, you have to design an amplifier for it, with surplus gain to be sacrificed and enough care over the phase shifts to keep it decently stable. The amplifier designed for feedback isn't much use without it. An amplifier not designed for feedback isn't much use with it. You either get too much gain or not enough.

David

[GrimJosef]

I think Leak normally worked with 26dB NFB, but given the transformers of the day and the high gain, so large number of stages, that he'd settled on, that was still pretty impressive. Audio designers did get better at it though. The Mullard 5-20 uses 30dB and they claim (although I think it may be a brave claim) that they've still got 10db of margin. The Brimar 25P1 power amp has 35dB of global NFB and an extra 5dB from some fiddly stuff going on in the small signal stages http://www.r-type.org/articles/art-101.htm.

Cheers,

GJ