Ouch.
There are umpteen ways to modify the design slightly, for slightly different results. My comments:
Firstly re the given schematic: The 0,22µF cap voltages as said. But I am not altogether certain why they are there in the first place; this is the first design I have seen with such.
Then, Ketje:
I do not understand why you say that the ECC83 phase inverter will not be able to drive EL34s satisfactorily if I understand you correctly? They seem to do quite well in the original original 5-20 circuit. (But I am not a fan myself of using such high-µ triodes for power stage drivers. My own prefernce would be for the EF86 pentode input tube with say ECC81 or ECC82 drivers for lower gain rather than the CB variation - but I understand his motivation for not trying to do too many changes. Still. But let us not come with too many variations.)
Also, I do not understand suggesting going to a triode input stage for lower sensitivity but then decreasing the global NFB increasing the sensitivity again. Also (sorry if I appear to be difficult - experience is a bugger ...) a proper UL output stage design has the advantage over both pentode and triode topology - not sure why alternatives to the EL34 UL configuration is suggested? But not wishing a derailment to a discussion of the relative merits of output configurations again; the data speaks for itself. Let's leave it at that.
There are umpteen ways to modify the design slightly, for slightly different results. My comments:
Firstly re the given schematic: The 0,22µF cap voltages as said. But I am not altogether certain why they are there in the first place; this is the first design I have seen with such.
Then, Ketje:
I do not understand why you say that the ECC83 phase inverter will not be able to drive EL34s satisfactorily if I understand you correctly? They seem to do quite well in the original original 5-20 circuit. (But I am not a fan myself of using such high-µ triodes for power stage drivers. My own prefernce would be for the EF86 pentode input tube with say ECC81 or ECC82 drivers for lower gain rather than the CB variation - but I understand his motivation for not trying to do too many changes. Still. But let us not come with too many variations.)
Also, I do not understand suggesting going to a triode input stage for lower sensitivity but then decreasing the global NFB increasing the sensitivity again. Also (sorry if I appear to be difficult - experience is a bugger ...) a proper UL output stage design has the advantage over both pentode and triode topology - not sure why alternatives to the EL34 UL configuration is suggested? But not wishing a derailment to a discussion of the relative merits of output configurations again; the data speaks for itself. Let's leave it at that.
Follow:
Ketje,
I just reread your post #3; saw that you also suggested rather an ECC82. I did not remember that - yes, that would do rather better than the ECC83 although the latter would suffice.
My apology.
Ketje,
I just reread your post #3; saw that you also suggested rather an ECC82. I did not remember that - yes, that would do rather better than the ECC83 although the latter would suffice.
My apology.
I asked Mr. Byrith about that, and this was his reply:
" The two condensers in connection with the Dc balance potmeter are there just to make sure that no AC or signal interferences can reach the control grids of the output valves. I know that if there were such spurious signals they would most certainly cancel because they would be in-phase on the grids where signals of opposite phase are the only ones passing through. Nevertheless it appears to me that it will always be better to make sure that only signals that are wanted can reach the control grids. I have once and only once had problems with spurious signals infested in the negative bias. But in my view two condenses are. a cheap insurance so that is the reason. Those who do not agree can just omit them."
He also mailed me an update on his previous documents about the Mullard "4-30", I attached it for those interested in his comments.
I am now convinced that I will be better off to just leave the design as is. As mentioned before here, I will not really save much by using smaller transformers etc.
I thought when I was in SA, it was difficult to source good quality components and the choices was not much. Here in the UK it is not much easier. To find a suitable HT transformer for example, I emailed 13 suppliers, 3 replied that they could not help me, I received 2 quotes from UK suppliers and 1 from the Netherlands. The rest simply did not bother to answer me.
RS electronics are still a good source, but again with limited choices. No good quality carbon composite resistors for example. They nowadays sell their own brand caps under "RS Pro". I ordered some for the HT supply, and noticed on receiving them, they were labeled "Jackcon". I immediately returned them and bought the 3 times more expensive Nichicons.
Now all the bits must arrive and then I can start building.
Thanks for all advice given here, appreciate.
Regarding the 0.22µF capacitors:
I am surprised that disturbing signals crept into the works at this late and insensitive stage in the gain chain, but if Mr Byrith found that, fine then. Also not a vast capital layout.
OT: Re components in the RSA:
As you said, though not all that pressing. I use MF resistors, moderately easy to come by as well as some reliable Philips and Japanese capacitors. But many of us import. Transformer wise: Several of us are capable of designing and winding our own, generally to be preferred in comparison to prohibitive import costs.
Good luck with your venture!
I am surprised that disturbing signals crept into the works at this late and insensitive stage in the gain chain, but if Mr Byrith found that, fine then. Also not a vast capital layout.
OT: Re components in the RSA:
As you said, though not all that pressing. I use MF resistors, moderately easy to come by as well as some reliable Philips and Japanese capacitors. But many of us import. Transformer wise: Several of us are capable of designing and winding our own, generally to be preferred in comparison to prohibitive import costs.
Good luck with your venture!
Surely it is simpler than that. They simply ensure that both halves of the phase splitter see identical loadings regardless of the positions of the two bias adjusting potentiometers. Otherwise if the bias was set towards one tube or the other there could be a corresponding difference in the amplitudes of the signals fed to the output tubes. With the pots at one extreme and no capacitors, here would be 390K on one side and 412K on the other as loadings on the phase splitter. With the capacitors, at audio frequencies, the loading is always 390K.
Regarding the setting of bias for different output tubes, yes.
But for the ECC83 total load the mentioned 390K and 412K values will be in parallel with the anode resistors of 150K plus whatever value the trimpot is set at. Computing that, the maximum difference between the load resistances for the triodes will be no more than 2% as a result of different grid resistances. The difference between the triodes in many ECC83s will be quite greater than that, thus the anode load trimpot. Then, and not to split hairs too finely, that is regarding the input signals to the power tubes, not taking any inequality in those themselves into account. (If one really wants to set the balance pots optimally within such small signal differences, distortion measurement at the OPT output will need to be the yardstick, not just signal equality at the grids.)
I read that Mr Byrith actually did experience unwanted signals (probably h.f.) entering the works, thus his desire to use those capacitors.
But for the ECC83 total load the mentioned 390K and 412K values will be in parallel with the anode resistors of 150K plus whatever value the trimpot is set at. Computing that, the maximum difference between the load resistances for the triodes will be no more than 2% as a result of different grid resistances. The difference between the triodes in many ECC83s will be quite greater than that, thus the anode load trimpot. Then, and not to split hairs too finely, that is regarding the input signals to the power tubes, not taking any inequality in those themselves into account. (If one really wants to set the balance pots optimally within such small signal differences, distortion measurement at the OPT output will need to be the yardstick, not just signal equality at the grids.)
I read that Mr Byrith actually did experience unwanted signals (probably h.f.) entering the works, thus his desire to use those capacitors.
I take your point - with the anode loads in the phase splitter the other differences become less significant.
"From the ox to the coat" (direct translation from an Afrikaans saying, meaning "onto a different subject" ), 😉 , looking at the 2 psu schematics attached, will it be ok to use OPTION 1, or should I stick to OPTION 2, which is the original CB design, with different values?
I would like to use OPTION 1 since I already have a dual capacitor of value 50+50uf 500V (C105A and C105B)
Will it matter if C104 and C105B are not the exact same values?
I would like to use OPTION 1 since I already have a dual capacitor of value 50+50uf 500V (C105A and C105B)
Will it matter if C104 and C105B are not the exact same values?
Attachments
Are there any? That means, isn't there an inherent antagonism between »good quality« and »carbon composite ressistors«?
Best regards!
Last edited:
Maybe because most certainly you won't find a tube manufaturer that makes tubes with their pins numbered ccw?
Best regards!
Best regards!
I finished the one amp (the mirrored one), and it is actually working. 😱
The mains supply is 240V here at my home, so my voltages are on the high side, hope it won't cause problems.
HT: 484V (supposed to be 450V)
The +410 is 447
The +160 is 202
Supply to EF86 is 92 (supposed to be 85)
Outputs from 12AX7 is 342 (supposed to be 320)
Heaters 6.5 (supposed to be 6.3)
I will adjust the resistors on the supplies to drop these to be more in line with the design.
I also included a switch to change between 12AX7 and 6N2P, if I want to. (Different heater configurations)
At the back are 3 testpoints for the DC balance and bias, the pots for these are accessible from above.
Here are some pics, nothing special, but it works and sounds GREAT. 😉
The mains supply is 240V here at my home, so my voltages are on the high side, hope it won't cause problems.
HT: 484V (supposed to be 450V)
The +410 is 447
The +160 is 202
Supply to EF86 is 92 (supposed to be 85)
Outputs from 12AX7 is 342 (supposed to be 320)
Heaters 6.5 (supposed to be 6.3)
I will adjust the resistors on the supplies to drop these to be more in line with the design.
I also included a switch to change between 12AX7 and 6N2P, if I want to. (Different heater configurations)
At the back are 3 testpoints for the DC balance and bias, the pots for these are accessible from above.
Here are some pics, nothing special, but it works and sounds GREAT. 😉
Attachments
Last edited:
Mirrored amps are either a waste of time, or worse. They may look nice, but mirroring guarantees that the two channels will behave slightly differently especially at higher frequencies. This because you cannot buy mirrored active devices so the stray capacitances and lead inductance will not be matched - unlike a pair of identical amps. Mirroring is for speakers, not amps.
JDeV,
I would not worry about the d.c. voltages so much as an increased heater supply. If your heater voltages measure above some 6,8Vac it may be prudent to insert a w.w. series resistor of some 47E in the mains lead - not a nice solution, but the only practical one which might be available to you. (Yah ..... I know some specs say up to 6,9V, but they do not specify what the norm is. Tube life will be shortened above some 6,6V depending.)
[We have the same problem here in RSA, and I reluctantly had to add a few-100 ohms in the mains leads of amplifiers now and then. Locally wound transformers have mains taps, allowing for up to 250V to be applied. And then there is mains regulation 🙁 .
I would not worry about the d.c. voltages so much as an increased heater supply. If your heater voltages measure above some 6,8Vac it may be prudent to insert a w.w. series resistor of some 47E in the mains lead - not a nice solution, but the only practical one which might be available to you. (Yah ..... I know some specs say up to 6,9V, but they do not specify what the norm is. Tube life will be shortened above some 6,6V depending.)
[We have the same problem here in RSA, and I reluctantly had to add a few-100 ohms in the mains leads of amplifiers now and then. Locally wound transformers have mains taps, allowing for up to 250V to be applied. And then there is mains regulation 🙁 .
I think it will take exactly the same amount of time to make the other one.
Who will REALLY notice the differance in behaviour without an oscilloscope, spectrum analyzer, distortion meter and a seismograph? I don't think I will and the symetrical appearance wil be more pleasing to me. ;-)
- Status
- Not open for further replies.