Wiring complete.
I finished the wiring today except for a bit of tidying-up. I have it running now for several hours with just the filaments connected without any HT supplies. Nevertheless, there is a lot of heat inside the chassis so I think a fan is essential. I've experimented with a 12 volt DC computer fan and have found that it works extremely well when fed from one of the 12 Volt AC supplies from T2, rectified by a single 1N4007 without any smoothing. It runs at about half speed without any vibration and very low noise.
I finished the wiring today except for a bit of tidying-up. I have it running now for several hours with just the filaments connected without any HT supplies. Nevertheless, there is a lot of heat inside the chassis so I think a fan is essential. I've experimented with a 12 volt DC computer fan and have found that it works extremely well when fed from one of the 12 Volt AC supplies from T2, rectified by a single 1N4007 without any smoothing. It runs at about half speed without any vibration and very low noise.
Attachments
Thanks for the encouragement. I'm now in the testing and measuring phase. All voltages and currents seem fine. It comes into balance perfectly after about 10 minutes warm up. I have one problem; I'm getting 50 - 60 volts on both outputs shortly after full switch-on. This lasts for about 15 seconds and the gas-discharge tubes are also flickering. Everything settles down and a reasonable balance is achieved shortly after that. I have experimented with and without a standby switch but it makes no difference. I cannot see myself having my loudspeakers connected to the outputs until I solve this problem. Can anybody throw some light on this?
Last edited:
What kind of voltage, dc or ac? 50-60v is very high in term of dc offset at any one time. Could be due to some oscillations at the beginning, try to reduce the 100k resistor in the output to 2.2k see what happens.
The voltage is DC and lasts for about 10 seconds. The gas-discharge tubes are on the point of firing but then the voltage quickly comes into balance. This is happening on both channels. I will try putting a low resistance, as you suggested and see what happens. Thanks.
I will do that if the lower resistance does not make a difference. I have a scope.
I think to be usefull to have some check points T1 Ua1,Ua2, Uk, T4 Ug and T5 Ug, also PSU HT1,2,3,4. Check also:
- if neon lamp is functioning
- if 60vdc varie uth the offset trimmer (1K) with imput shorted
- if bias trimmer is functionnig (200mA accross a 0,1resistor with imput shorted)
- if C5 has +to ground
- if neon lamp is functioning
- if 60vdc varie uth the offset trimmer (1K) with imput shorted
- if bias trimmer is functionnig (200mA accross a 0,1resistor with imput shorted)
- if C5 has +to ground
Thanks for all those suggestions, Victor. In the meantime I have found that putting a 1.5 K resistor across the outputs reduces the initial voltage offset to about 5 Volts. That's a big drop from what turned out to be 80 Volts. I must check it's polarity. It's strange that the problem is identical on both channels. I've just realized, from what you said that I'm running the amp without the inputs shorted. Anyway the oscilloscope is next.
Be shure, the values in the amp, as designed by Tim, are perfect. So I think the problem can be also in the PSU. I suggest not to alter the original values. Make adjustments only with imput shorted.
Offset must not esceed 30mV (mine show 10mV)
Offset must not esceed 30mV (mine show 10mV)
By the way, what kind of load did you use? I suggest 8ohms/50W Arcol. Do not power on the amp without one.
Ok, well, I have to admit that I was running it with no load. I did this on the understanding that if there was close to 0 Volts on the output, there would not need to be a load for testing. I'll keep reducing the resistor on the outputs to see what happens.
OK, with dummy loads of 39 Ohms on both channels, the initial offset shortly after full switch-on is only 100 milliVolts with or without the inputs shorted.
You can also swap the top tube with lower one to confirm that the offset is really due to difference in the time for the tubes to stabilize, I remember some tubes take much longer, So you need to check the offset after every half hour. When you swap the tubes, the polarity could be reversed (?). Double confirm by checking voltage at R1 (AC) that also reflects the amount of offset.
It looks like I have a bum EF86 in the four I bought as a matched quad. The problem of one channel taking over 30 minutes to stabilize travels with the tube from side to side in either V2 or V3 positions. The channel without the faulty EF86 always comes into balance within 30 seconds.
I see a lot of Svetlana equivalents from Russia on eBay very cheap. The data sheet says they have a gm of 1.8 +/- 0.5 which is a bit less than the EF 86 at 2.2. I'm thinking of buying 10 of them and matching them myself. Any comments?
So matched gm (AC wise) is no problem, what about DC condition for the bias it could be different for gm matched pairs, beside you also need AC/DC matched pairs in the front-end, that can compound the problem, So you end up matched pairs working in unmatched situation. I think overall gain is less important as you have volume control to adjust the output level, but to get the AC balanced is important as you don't want too much DC offset at higher output. And this is determined by all the tubes working together, so I guess you just have to try it using unmatched tube targeting for least DC offset.
I think Koonw is right. Try diffrent ef86 till you get what you need (low offset as well as balanced +/- voltage) The ef86 work together with asociated 6c33c. As you can hardly match the 2x6c33c you shal trick them by matching with the asociated ef86. I have allready done this. I suggest also RFT or TESLA NOS ef86 against new russian ones. Also "cooking" at least 24 hours the tubes is a wise step.