Hi,
Parafeed cap should go to the top of the cathode bias resistor of the bottom 807 for best results.
Said cathode resistor should be sufficiently capacitively bypassed.
Otherwise it won't make any difference whether the parafeed cap is sitting at the B+ side or ground side of the SE OPT.
The point is to impede DC flow; either works.
In the case of the mu-follower I'd recommend putting the caps as I suggested....It makes for a clearly measureable difference.
Cheers, 😉
Parafeed cap should go to the top of the cathode bias resistor of the bottom 807 for best results.
Said cathode resistor should be sufficiently capacitively bypassed.
Otherwise it won't make any difference whether the parafeed cap is sitting at the B+ side or ground side of the SE OPT.
The point is to impede DC flow; either works.
In the case of the mu-follower I'd recommend putting the caps as I suggested....It makes for a clearly measureable difference.
Cheers, 😉
Placing the cap at the top was in particular reference to the stock Paramour output transformer, as the article said that it wasn't designed to take the full B+. This is where I got the idea of safety, my transformers are designed for high voltage so that shouldn't be an issue. I will have to ground the transformer core and the speaker return terminal to address general safety concerns.
So the feeling is that the Parafeed cap should be returned to the cathode of the 807 rather than at the anode. I can well understand the reason why it might have a sonic impact on the result, and thats why I originally placed it at the bottom.
Its relatively easy for me to bypass the cathode resistor. One of Gary Pimms articles sujests that the cathode bypass cap isn't needed in the parafeed arrangement, but I think this was more particularly relevant to preamp application.I would anticipate the voltage swing on the cathode to be about +40V max, so quality 63V caps should do the job. In the original RH807SE they are rated at 100V, but I put that down to the usual audio designers caution in these matters.
So the feeling is that the Parafeed cap should be returned to the cathode of the 807 rather than at the anode. I can well understand the reason why it might have a sonic impact on the result, and thats why I originally placed it at the bottom.
Its relatively easy for me to bypass the cathode resistor. One of Gary Pimms articles sujests that the cathode bypass cap isn't needed in the parafeed arrangement, but I think this was more particularly relevant to preamp application.I would anticipate the voltage swing on the cathode to be about +40V max, so quality 63V caps should do the job. In the original RH807SE they are rated at 100V, but I put that down to the usual audio designers caution in these matters.
Hi,
I don't think so.
Don't forget you have a DC voltage on the cathode as well, add your AC swing and 100VDC insulation voltage for a cathode bypass cap doesn't look like being cautious anymore.
Output tubes' cathodes should always be properly bypassed to maintain bias stability.
Putting the parafeed cap and bypass cap the way I suggest will make the negative effects of the, usually, electrolytic cathode bypass cap pretty much inaudible.
Give it a try and let us know what you think....
Cheers, 😉
I don't think so.
Don't forget you have a DC voltage on the cathode as well, add your AC swing and 100VDC insulation voltage for a cathode bypass cap doesn't look like being cautious anymore.
Output tubes' cathodes should always be properly bypassed to maintain bias stability.
Putting the parafeed cap and bypass cap the way I suggest will make the negative effects of the, usually, electrolytic cathode bypass cap pretty much inaudible.
Give it a try and let us know what you think....
Cheers, 😉
Thanks for the help.
All I've got to do now is build the thing.
I'll keep you all posted on progress.
Shoog
All I've got to do now is build the thing.
I'll keep you all posted on progress.
Shoog
Having just read the Tube CAD article on partial feedback, I am having real concerns about whether this design approach will mean that my 4uf paper in oil caps will just be too small for the parafeed duty. The output impedence will be really low (lower than a triode I think) and the 4uf cap was marginal for a normal triode parafeed design.
However if I drop the partial feedback and go for a plain tetrode design, it will undermine the effectiveness of the CCS, and probably mean that the circuit won't work anyway.
I just wouldn't be able to live with the expected 1.8W of a triode strapped 807.
Any thoughts !!
Shoog
However if I drop the partial feedback and go for a plain tetrode design, it will undermine the effectiveness of the CCS, and probably mean that the circuit won't work anyway.
I just wouldn't be able to live with the expected 1.8W of a triode strapped 807.
Any thoughts !!
Shoog
Shoog said:
Maybe I'm missing something here. The lower plate resistance allows the use of a smaller parafeed cap. So, if you now have lower effective rp because of the partial feedback, your current cap should be fine. Parafeed is part black art anyway, and you usually need to experiment with different cap sizes anyway - so build it and see. It will probably be fine.
I thought that the size of cap was defined by the difference in impedance on either side of the cap, so a lower rp whilst maintaining the plate to plate impedance would increase the parafeed cap required. Isn't it the case in a similar situation with cathode follower preamp stages needing a larger coupling cap than an anode follower, which has a a higher output impedance.
Shoog
Shoog
Shoog said:
you're right
I was thinking of rp vs OPT inductance so...
C=1/2(pi)*F3*rp (Voltsecond's method)
... unless your new rp is an order of magnitude lower than before, you'll still be in the ball park. Try it and see. You can always switch caps.
Thanks for the reasurance and the formula. I worked it out that even with the expected rp of 192R, I would still only need a 1uf cap to bring the resonance down to about 20Hz. Sounds a bit dubious to me, but I will build and see.
I know what you mean about the black arts of parafeed capacitors, ask any two people for a value and they will come up with wildly different answers.
We will see.
Shoog
I know what you mean about the black arts of parafeed capacitors, ask any two people for a value and they will come up with wildly different answers.
We will see.
Shoog
Hi again,
Just a status report on the amp, and a question.
I have just put together the power supply. The main HT supply which needs to be a 750V was modelled in PSU Designer II. It has a 900-0-900V secondary supplying a 5R4. The filtering is 30H 32R choke followed by 3uf followed by 30H 32R choke followed by 4uf . There is then a bleed resistor of 30K, which should present a load of about 33mA. Modelling suggests an output DC of about 780V which is about right for the circuit. However the actual voltage given by the supply is 970V. A bit of a discrepancy, requiring a complete redesign to use. I have a feeling that even under the expected full load of 133mA this supply will barely sag at all. The transformer is designed to give out 450mA.
There are a few other problems as well, the first choke buzzes quite noticably, it is designed for this role so I suppose I will just have to live with it (unless anyone can sujest otherwise). The 5R4 has flash over on start up, which again is to be expected at this much of an over voltage, if I can get the voltage down then it should disappear.
My suggested solution for this is to put resistance in the secondaries before the 5R4. This will hopefully drop the voltage by the desired amount, take the strain off the 5R4. I was thinking of using two 24W 1K resistors. Would I be correct in assuming that each will handle half of the final 133mA of load, which should keep them within ratings. Are there any other issues I should be aware of with this approach.
Thanks
Shoog
Just a status report on the amp, and a question.
I have just put together the power supply. The main HT supply which needs to be a 750V was modelled in PSU Designer II. It has a 900-0-900V secondary supplying a 5R4. The filtering is 30H 32R choke followed by 3uf followed by 30H 32R choke followed by 4uf . There is then a bleed resistor of 30K, which should present a load of about 33mA. Modelling suggests an output DC of about 780V which is about right for the circuit. However the actual voltage given by the supply is 970V. A bit of a discrepancy, requiring a complete redesign to use. I have a feeling that even under the expected full load of 133mA this supply will barely sag at all. The transformer is designed to give out 450mA.
There are a few other problems as well, the first choke buzzes quite noticably, it is designed for this role so I suppose I will just have to live with it (unless anyone can sujest otherwise). The 5R4 has flash over on start up, which again is to be expected at this much of an over voltage, if I can get the voltage down then it should disappear.
My suggested solution for this is to put resistance in the secondaries before the 5R4. This will hopefully drop the voltage by the desired amount, take the strain off the 5R4. I was thinking of using two 24W 1K resistors. Would I be correct in assuming that each will handle half of the final 133mA of load, which should keep them within ratings. Are there any other issues I should be aware of with this approach.
Thanks
Shoog
Shoog said:
Hi there....... the 5R4 is spec'd for 750-0-750 V.....(unless I'm wrong)....it seems you are operating it under murderous conditions.....no wonder flashover.
As to the buzzing choke.....I had one and I dunked it in a pot of polyeth wood varnish..I was lucky and that shut it up. If the loose lamination is in the middle of core, short of splitting the iron.....you'll have to live with it or bin it.
With the resistor approach in each secondary tranny leg (it does seem a pity of the HV overwind) you won't get far with that in AB. Using the higher volt taps on the primary is a way out but won't rescue the situation as the rect heater volts will fall. If there is a gap between the laminations and bobbin one can slip in an extra few turns round the bobbin to make up the volts loss.......but I wouldn't advise anyone doing that approach if one hasn't had experience of mains trannies.........
richj
"With the resistor approach in each secondary tranny leg (it does seem a pity of the HV overwind) you won't get far with that in AB."
The circuit is pure class A with CCS suppying the output valves. As such it should work in this application. My calculation is that the 1K in each leg will give me a drop of about 250V, which will be a little lower than my designed B+ of 750V (970-250=720V approximately). The option of dropping the primary voltage isn't an option because its already on its highest tapping.
I will try the other choke in the first position just to see if it behaves any better.
Thanks Shoog
The circuit is pure class A with CCS suppying the output valves. As such it should work in this application. My calculation is that the 1K in each leg will give me a drop of about 250V, which will be a little lower than my designed B+ of 750V (970-250=720V approximately). The option of dropping the primary voltage isn't an option because its already on its highest tapping.
I will try the other choke in the first position just to see if it behaves any better.
Thanks Shoog
Todays experiement has produced the following result. Placing a 1K resistor in each side of the transformer secondary has produced an AC voltage drop of 20V AC, with a load of 20mA through the resistor. This translates to an 80V AC drop with the full load of 133mA. This only produces a DC drop of 20V after rectification, but the addition of a 1K resistor after the first choke should drop about 120V DC.
This means that with the 1K resistors in the legs of the secondary, and a 1K resistor in the supply, I should get 200V DC drop, which will leave me at about 770V DC, which is just about exactly where i want to be.
The resistors in the secondary legs have removed the flash over, and the choke is a lot quieter.
Shoog
This means that with the 1K resistors in the legs of the secondary, and a 1K resistor in the supply, I should get 200V DC drop, which will leave me at about 770V DC, which is just about exactly where i want to be.
The resistors in the secondary legs have removed the flash over, and the choke is a lot quieter.
Shoog
Hi there,
Latest status report.
I have finished the initial build, and after a few adjustments to get exactly the correct voltages, it works - sort of.
Music came out of both channels first time, and nothing blew up. I have tested it with a pair of test speakers and the headphones output of an old radio.
The only serious issue initially is that it has an instability problem. It motorboats at about 50Hz (this is not mains hum as sometimes it completely absent), it also changes it characture over time. This is also distorting the signal. Signal seems to effect its presence or absense. Could potential be pickup from the mains transformer, as the amp is nude on the bench. Could also be grounding issues. Could also be instability in the output transformers (I am using 8V mains transformers at the moment, and half of the winding is unterminated).
The TT21 sujests using a bit of resistance on the top cap, but this could be technically messy to add.
Will think on this and see if anything obvious sujests itself.
Shoog
Latest status report.
I have finished the initial build, and after a few adjustments to get exactly the correct voltages, it works - sort of.
Music came out of both channels first time, and nothing blew up. I have tested it with a pair of test speakers and the headphones output of an old radio.
The only serious issue initially is that it has an instability problem. It motorboats at about 50Hz (this is not mains hum as sometimes it completely absent), it also changes it characture over time. This is also distorting the signal. Signal seems to effect its presence or absense. Could potential be pickup from the mains transformer, as the amp is nude on the bench. Could also be grounding issues. Could also be instability in the output transformers (I am using 8V mains transformers at the moment, and half of the winding is unterminated).
The TT21 sujests using a bit of resistance on the top cap, but this could be technically messy to add.
Will think on this and see if anything obvious sujests itself.
Shoog
Hi there,
Just rewired the output transformers, referencing them to earth. Motoboating mostly gone. I would say that its probably the radio picking up tranformer hum. Without any input there is a slight hum and a slight hiss. The hum would be mains pickup, and can probably be eliminated by looking at the earthing.
Scoping up the output shows it fairly clean. Little difficult to say how its performing until I put my signal generator through it.
Used Gary Pimms pentode CCS with the TT21's and this seems to be performing well.
The bass seems to be reasonably well extended, a little difficult to tell until I put it through my main system.
All in all its been remarkably painless to get to this stage.
Shoog
Just rewired the output transformers, referencing them to earth. Motoboating mostly gone. I would say that its probably the radio picking up tranformer hum. Without any input there is a slight hum and a slight hiss. The hum would be mains pickup, and can probably be eliminated by looking at the earthing.
Scoping up the output shows it fairly clean. Little difficult to say how its performing until I put my signal generator through it.
Used Gary Pimms pentode CCS with the TT21's and this seems to be performing well.
The bass seems to be reasonably well extended, a little difficult to tell until I put it through my main system.
All in all its been remarkably painless to get to this stage.
Shoog
The motorboating is still there, but intermittent.
Replaced the 8uf cap in the screen/input triode supply with a 330uf cap. Thought this would probably sort the motorboating out, as interaction between stages is a common cause, hasn't worked.
Discovered a really fundamental mistake. With one channel input there is still output from both speakers. Must have crossed over one of the wires between channels. This could well be the source of the motorboating. Will track this fault down before doing anything else. Its really tight in the circuit so i'am not really looking forward to this at all.
Tried two different ECC81's, one a Mullard, the other unbranded. makes a significant difference to the tone. The Mullard is much mellower.
Happy days.
Shoog
Replaced the 8uf cap in the screen/input triode supply with a 330uf cap. Thought this would probably sort the motorboating out, as interaction between stages is a common cause, hasn't worked.
Discovered a really fundamental mistake. With one channel input there is still output from both speakers. Must have crossed over one of the wires between channels. This could well be the source of the motorboating. Will track this fault down before doing anything else. Its really tight in the circuit so i'am not really looking forward to this at all.
Tried two different ECC81's, one a Mullard, the other unbranded. makes a significant difference to the tone. The Mullard is much mellower.
Happy days.
Shoog
Sorted out the crossed over channels. Thats OK.
I have isolated the motoboating componet, seems to be about 30Hz with a saw tooth shape with a kink where it crosses over the voltage zero point. Everything is superimposed on this.
There is also another oscillation component, with a period of about 5us (100Khz I think). This is a clean sine wave and rides on the audio signal without varying.
I tried placing a 100ohm resistor after the final cap in the main power supply, didn't have any effect.
The main motorboating doesn't come in until the whole amp is nicely warmed up, and then it just suddenly kicks in.
Maybe its the transitorised screen supply of the CCS starving the screen.
I tried using some small ceramic caps across the output transformer secondaries, but to no noticable effect.
I know the theory of what might be causing it but its real difficult to identify its source. I can't monitor a lot of the circuit because my scope just can't cope with the voltages present in the CCS.
Any ideas.
Shoog
I have isolated the motoboating componet, seems to be about 30Hz with a saw tooth shape with a kink where it crosses over the voltage zero point. Everything is superimposed on this.
There is also another oscillation component, with a period of about 5us (100Khz I think). This is a clean sine wave and rides on the audio signal without varying.
I tried placing a 100ohm resistor after the final cap in the main power supply, didn't have any effect.
The main motorboating doesn't come in until the whole amp is nicely warmed up, and then it just suddenly kicks in.
Maybe its the transitorised screen supply of the CCS starving the screen.
I tried using some small ceramic caps across the output transformer secondaries, but to no noticable effect.
I know the theory of what might be causing it but its real difficult to identify its source. I can't monitor a lot of the circuit because my scope just can't cope with the voltages present in the CCS.
Any ideas.
Shoog
Got the thing stable. A few crossed over wires had turned the output tubes into a nice stable oscillator.
Only problem I now have is that the HT line just doesn't have anything like enough final capacitance. This means that when it is playing at anything over a very quiet, it breaks up in a very bad way. What happens is that it sort of pops and the signal bleeds between the channels. I imaging that its the |CCS complaining and giving up the challenge.
There are two potential solutions to my way of thinking. Add lots of very expensive and difficult to source 1000V caps, or alternatively look at the options with regards to regulating the HT supply.
Keep you all posted.
Shoog
Only problem I now have is that the HT line just doesn't have anything like enough final capacitance. This means that when it is playing at anything over a very quiet, it breaks up in a very bad way. What happens is that it sort of pops and the signal bleeds between the channels. I imaging that its the |CCS complaining and giving up the challenge.
There are two potential solutions to my way of thinking. Add lots of very expensive and difficult to source 1000V caps, or alternatively look at the options with regards to regulating the HT supply.
Keep you all posted.
Shoog
Did some more messing about today. Tried adding a stack of 200V caps to give me an extra 110uf of filtering.
However, I managed to get my scope to monitor the HT rail, and surprise, surprise it very flat and stable. This therefore is not the source of the problem.
Thought the poping sound was probably the sound of the voltage reference switching itself on and off, as scoping the output shows it jumping up and down with each pop. Tried reducing the resistors supplying the voltage reference TL1431 in order to give it at least 2mA of bias current. Didn't seem to help much. Next I measured some voltages. The current source isn't behaving as expected. I designed it to drop 400V and at idle its barely dropping 200V. With a signal it drops 300V. I decided to try decreasing the reference voltage from 36V to 27V, seems to have reduced the popping on the channel I changed, but still noisy and distorting. It seems to be working after a fashion as the 807 has a bias of 18V over its 400R cathode resistor, which indicates it should be dropping about 45mA (as expected). The 807 can't live long with nearly 600V on it, but the things I have tried so far don't force the TT21 to use more voltage. It could be that the TT21 is generating the noise and popping because its running at to low a voltage.
Maybe the TT21 and the 807 are generating a voltage divider because of there plate resistances. Bypassing the CCS to reduce its plate resistance could work. I'am guessing at this stage.
Shoog
However, I managed to get my scope to monitor the HT rail, and surprise, surprise it very flat and stable. This therefore is not the source of the problem.
Thought the poping sound was probably the sound of the voltage reference switching itself on and off, as scoping the output shows it jumping up and down with each pop. Tried reducing the resistors supplying the voltage reference TL1431 in order to give it at least 2mA of bias current. Didn't seem to help much. Next I measured some voltages. The current source isn't behaving as expected. I designed it to drop 400V and at idle its barely dropping 200V. With a signal it drops 300V. I decided to try decreasing the reference voltage from 36V to 27V, seems to have reduced the popping on the channel I changed, but still noisy and distorting. It seems to be working after a fashion as the 807 has a bias of 18V over its 400R cathode resistor, which indicates it should be dropping about 45mA (as expected). The 807 can't live long with nearly 600V on it, but the things I have tried so far don't force the TT21 to use more voltage. It could be that the TT21 is generating the noise and popping because its running at to low a voltage.
Maybe the TT21 and the 807 are generating a voltage divider because of there plate resistances. Bypassing the CCS to reduce its plate resistance could work. I'am guessing at this stage.
Shoog
Progress has been made.
I checked the 807 screen supply voltage and it was 276V. I had designed it to be 250V which makes about 20mA difference. The result was that the 807 were going hungry. After adjusting the screen supply the distortion and the popping have disapeared.
Sounds very nice.
Unfortunately the voltage on the plate of the 807 has now stablised at 627V ( I had hoped that the screen adjustment would reduce the voltage). The CCS just doesn't seem to want any more than 150V across it. I need to get this up to 400V, any ideas??
I checked the 807 screen supply voltage and it was 276V. I had designed it to be 250V which makes about 20mA difference. The result was that the 807 were going hungry. After adjusting the screen supply the distortion and the popping have disapeared.
Sounds very nice.
Unfortunately the voltage on the plate of the 807 has now stablised at 627V ( I had hoped that the screen adjustment would reduce the voltage). The CCS just doesn't seem to want any more than 150V across it. I need to get this up to 400V, any ideas??
- Status
- Not open for further replies.