I have on the test bench a jig to test this. Depletion MOS or a resistor, LED or bypassed cathode resistor, or direct DC bias on the grid.
Merlin Blencowe's paper "Noise in Triodes with Particular Reference to Phono Preamplifiers" describes an optimal current at which a noise minima exists. Unfortunately, all his tests were performed on 9-pin tubes. I am going to guess that 8mA is a bit "hot" and needs to be dialed back.
The quietest 6SN7's I have were Tung Sols obtained from a fellow in Canada who had salvaged them! Usually older tubes get noisier with age.
Hi Jac,
Those are insulated gate mosfets and so are static sensitive. The source-gate voltage probably has an abs. max. of about 20 VDC. Taken beyond this point you can easily have some breakdown of the oxide and increased leakage current (which might be your issue here). I normally use these as current sources and haven't any problems with any I have bought over the years, even the new ones I just bought. Maybe I will, maybe not. Microchip is usually a very good manufacturer, so they would be the last place I would be looking for those problems.
As for input impedance to amplifiers, tube product was commonly 500K down to a low of about 100K. A 1 meg resistor from the input tube grid to ground is often 1 megohm, down to maybe 470 Kohm. Solid state folks have a hard time thinking about resistances that high, so what Stuart did was very normal and correct. Of course, you would be using low capacitance wire for any tube equipment, not just this preamp.
I hope that helps.
-Chris
Those are insulated gate mosfets and so are static sensitive. The source-gate voltage probably has an abs. max. of about 20 VDC. Taken beyond this point you can easily have some breakdown of the oxide and increased leakage current (which might be your issue here). I normally use these as current sources and haven't any problems with any I have bought over the years, even the new ones I just bought. Maybe I will, maybe not. Microchip is usually a very good manufacturer, so they would be the last place I would be looking for those problems.
As for input impedance to amplifiers, tube product was commonly 500K down to a low of about 100K. A 1 meg resistor from the input tube grid to ground is often 1 megohm, down to maybe 470 Kohm. Solid state folks have a hard time thinking about resistances that high, so what Stuart did was very normal and correct. Of course, you would be using low capacitance wire for any tube equipment, not just this preamp.
I hope that helps.
-Chris
Hi Chris,
Thanks for your response. That's the good thing about these forums. I always learn something new.
Regarding the high resistance values for the tube, you are correct in that I have learned about electronics through solid state and am just learning the mysteries of tubes. You mentioned that it is common to have high resistances and that SY's choices were normal. Is it just tradition or is there some functional reason for the high resistances? I can see that tubes deal with much higher voltages and, often, lower currents which seems to imply limiting current through high resistance would be a good idea in case of a failure. I'm just trying to wrap my head around it.
Are there reasons that I shouldn't try lower resistances (22k for example) on the input impedance resistor? At present, the main thing I am optimizing is bandwidth, but maybe I am missing a bigger tradeoff. It looks like I need to dig deeper into Morgan Jones.
As for Microchip, I agree that they are usually an excellent manufacturer. In my first issue with the EO phono preamp, I assumed that I did something to damage the transistor and create the leak, possibly static charge or overheat in soldering. This time, I was very careful of static charge protocol and tested them as soon as I got them. I also use a 9V battery and pot to control the source-gate voltage to avoid pushing that 20V limit. It's possible that I am making some other testing mistake, but I am pretty certain that the leaky parts came to me that way. You are right in that there are too many ways for me to screw up the transistor and I shouldn't be blaming Microchip. What think I can say is that leaky transistors in a CCS can result in noise of the type that Mr. Dave was experiencing and checking your CCS mosfets for leakage is probably a good idea, if you are having noise problems.
Jac
Thanks for your response. That's the good thing about these forums. I always learn something new.
Regarding the high resistance values for the tube, you are correct in that I have learned about electronics through solid state and am just learning the mysteries of tubes. You mentioned that it is common to have high resistances and that SY's choices were normal. Is it just tradition or is there some functional reason for the high resistances? I can see that tubes deal with much higher voltages and, often, lower currents which seems to imply limiting current through high resistance would be a good idea in case of a failure. I'm just trying to wrap my head around it.
Are there reasons that I shouldn't try lower resistances (22k for example) on the input impedance resistor? At present, the main thing I am optimizing is bandwidth, but maybe I am missing a bigger tradeoff. It looks like I need to dig deeper into Morgan Jones.
As for Microchip, I agree that they are usually an excellent manufacturer. In my first issue with the EO phono preamp, I assumed that I did something to damage the transistor and create the leak, possibly static charge or overheat in soldering. This time, I was very careful of static charge protocol and tested them as soon as I got them. I also use a 9V battery and pot to control the source-gate voltage to avoid pushing that 20V limit. It's possible that I am making some other testing mistake, but I am pretty certain that the leaky parts came to me that way. You are right in that there are too many ways for me to screw up the transistor and I shouldn't be blaming Microchip. What think I can say is that leaky transistors in a CCS can result in noise of the type that Mr. Dave was experiencing and checking your CCS mosfets for leakage is probably a good idea, if you are having noise problems.
Jac
Input Impedance and input zobel
A little more careful reading of SY's AX article and I think I have realized my error. The key is that SY used a 15k resistor with no capacitor in the zobel and that this is in parallel with the 100k pot and 1 Meg resistor. That means that the input impedance seen by the trafo is much lower than I had been calculating because I was ignoring the zobel. As SY built his Impasse, input impedance would be 11.5k at mid-volume and 13k at zero volume. As he said in the article, "You need to pick the input circuitry as a team...... You must first cater to the whims of the prima donna, the input transformer." What that means is that SY is right per usual and you guys who used a 100k pot and a zobel resistor in the 10 to 15k range are doing just fine.
And, I've learned a little bit.
Jac
A little more careful reading of SY's AX article and I think I have realized my error. The key is that SY used a 15k resistor with no capacitor in the zobel and that this is in parallel with the 100k pot and 1 Meg resistor. That means that the input impedance seen by the trafo is much lower than I had been calculating because I was ignoring the zobel. As SY built his Impasse, input impedance would be 11.5k at mid-volume and 13k at zero volume. As he said in the article, "You need to pick the input circuitry as a team...... You must first cater to the whims of the prima donna, the input transformer." What that means is that SY is right per usual and you guys who used a 100k pot and a zobel resistor in the 10 to 15k range are doing just fine.
And, I've learned a little bit.
Jac
Hi Jac,
Stuart says the 1M0 resistor is for safety in case the wiper of the control goes open. That way the tube will always have a DC connection to ground. I assumed there was a capacitor between the wiper and the tube, which would require this resistor absolutely. The way Stuart designed the preamp, it is a safety feature only.
Yes, most transformer inputs would require a matching network of some kind to avoid problems with leakage capacitance and inductance. You surmised correctly that the zobel and resistor were loading the transformer so it was "happy" with the load impedance.
Best, Chris
Stuart says the 1M0 resistor is for safety in case the wiper of the control goes open. That way the tube will always have a DC connection to ground. I assumed there was a capacitor between the wiper and the tube, which would require this resistor absolutely. The way Stuart designed the preamp, it is a safety feature only.
Yes, most transformer inputs would require a matching network of some kind to avoid problems with leakage capacitance and inductance. You surmised correctly that the zobel and resistor were loading the transformer so it was "happy" with the load impedance.
Best, Chris
Hi,
Am still having issues. I have tried a third set of mosfets. This time i was much more careful regarding static. I now have a mat! Still the intermittant fizzing which is quite quiet. I did build the test circuit on a bread board, any static issues with those? The only other thing (if not what you suggest) is may adjustment resistors. I have been using some somewhat cheap metal films. But i would think that either, theyre noisy full stop or would properly fail. But i dont think R5 is being stressed at all. Also, the noise didnt seem any worse than with the stock resistors that came with the kit. I have upgraded almost all the other resistors in the circuit with prp (the red ones from hifi collective).
Three different lots of fets.
Some new behaviour has happened though.
A few nights ago all of a sudden, quite a lot of noise was coming through the right channel. This stayed with the volume control down. Following morning tried again and same.
I come home last night and the Mrs is watching tele and its working fine (well, with the usual occasional tiny fizzle).
Sometimes turning it off and on again seems to reduce the problem.
I have been wondering whether i damaged the pass transistor or the lm317 when i had my mishap when building the output board. As i have rebuilt almost everything on the output board i am running out of options. But presumably if it were the power supply, the noise should show up on both channels, there may be more rejection (something to do with the variance in plate resistance maybe?) on one side but surely the same noise, even if at different levels should be present on both.
It does sound like it likely to be the issue you describe as with each rebuild the noise seems to change.
Was your fizzing sound constant or intermittant?
Thank you for this info. Its really doing my head in and i now feel considerably closer!
Hi Dave,
I understand your frustration. When I can't find a problem like that, I almost get desperate.
To answer your question, my noise was constant and any leaky transistors that I found have been able to be measured consistently as leaky. That may mean that my problem isn't your problem.
As for resistors, I have never seen a noise issue with one. That said, several people have told me they have lots of experience with the DN2540 mosfets and no bad ones, so who can say. I also agree that R5 isn't really under any significant current or stress.
The intermittent behavior of your device suggests something is unstable or a resistance is changing over time. I'm sure you have triple checked for cold solder joints. Is it possible that you have a mechanical connection to some device that is an intermittent connection ar variable resistance? Being careful to be insulated from high voltage, I might be tempted to wiggle things to see if the noise occurs. It is always possible that a joint was mechanically damaged when bending a lead or something.
One last crazy idea. I have a solid state amp that I built that makes a noise like yours when my mobile phone is on and nearby the amp. The rest of the time, it is quiet. Apparently, there is too big a hole somewhere in my case and radio frequency noise is getting in. Could there be an intermittent radio source causing your problem? I have to admit, once I figured out it was the mobile phone, I stopped worrying about it completely.
I sure hope other guys join in and help you find it. I'm certainly no expert and I'm sure others will have good ideas.
Jac
I understand your frustration. When I can't find a problem like that, I almost get desperate.
To answer your question, my noise was constant and any leaky transistors that I found have been able to be measured consistently as leaky. That may mean that my problem isn't your problem.
As for resistors, I have never seen a noise issue with one. That said, several people have told me they have lots of experience with the DN2540 mosfets and no bad ones, so who can say. I also agree that R5 isn't really under any significant current or stress.
The intermittent behavior of your device suggests something is unstable or a resistance is changing over time. I'm sure you have triple checked for cold solder joints. Is it possible that you have a mechanical connection to some device that is an intermittent connection ar variable resistance? Being careful to be insulated from high voltage, I might be tempted to wiggle things to see if the noise occurs. It is always possible that a joint was mechanically damaged when bending a lead or something.
One last crazy idea. I have a solid state amp that I built that makes a noise like yours when my mobile phone is on and nearby the amp. The rest of the time, it is quiet. Apparently, there is too big a hole somewhere in my case and radio frequency noise is getting in. Could there be an intermittent radio source causing your problem? I have to admit, once I figured out it was the mobile phone, I stopped worrying about it completely.
I sure hope other guys join in and help you find it. I'm certainly no expert and I'm sure others will have good ideas.
Jac
Exactly! My values were a little different, but it should work. I used a 9V battery with a pot to vary the negative voltage on the gate to source. I guess that worked like the gate resistor. I don't know enough to know if you can eliminate it, but it shouldn't matter.
Jac
I have poked around with a wooden chopstick but currently to no avail.
I have also wondered whether its severe mains borne interference. Although i have filtered and upgraded mains cables with ferrites etc etc. Tried deplugging essentially everything, turned fridges off etc.
Sadly my scope seems to be just too noisy to be of real use in these kind of situations. I will take another probe about though (and probably swap out the LM317 anyway as i have a stack of those).
The phone thing has occurred to me. But doesnt seem to be quite consistent enough to warrant noting a correlation. I haven't ruled it out. But some of my other kit (particularly an old pair of krk studio monitors that i use in the office via unbalanced cable) does seem very susceptible to that " ber dat bah , ber dat bah , " sound that you get, particularly with guitar amps and the like. But don't get that at all through the hifi.
I think the case is pretty well screened. There are some slots on the top for ventilation but they are quite narrow. Its one of the 10mm aluminum fronted steel cases from hifi2000.
I think ill have a play around and just see whether any of these remaining fets I've got are duff and maybe see what it take to break one (static carelessness wise) to see if thats likely to have happened to the ones in there.
Actually something has just occured to me. Can tv coax be leaky if its cheap? Thats something i havent tried unplugging from the wall, and that does run behind all the stereo equipment, probably getting jostled about everytime i remove the thing to take to the test bench.
I think its unlikely as the problem follows the preamp channel rather than the cables.
I have certainly thought this fizzing sound has a slightly digital sounding nature to it which is why i removed my mains networking sockets in an attempt to clear it up (it didnt, but i no longer use them anyway).
Cheers
I have also wondered whether its severe mains borne interference. Although i have filtered and upgraded mains cables with ferrites etc etc. Tried deplugging essentially everything, turned fridges off etc.
Sadly my scope seems to be just too noisy to be of real use in these kind of situations. I will take another probe about though (and probably swap out the LM317 anyway as i have a stack of those).
The phone thing has occurred to me. But doesnt seem to be quite consistent enough to warrant noting a correlation. I haven't ruled it out. But some of my other kit (particularly an old pair of krk studio monitors that i use in the office via unbalanced cable) does seem very susceptible to that " ber dat bah , ber dat bah , " sound that you get, particularly with guitar amps and the like. But don't get that at all through the hifi.
I think the case is pretty well screened. There are some slots on the top for ventilation but they are quite narrow. Its one of the 10mm aluminum fronted steel cases from hifi2000.
I think ill have a play around and just see whether any of these remaining fets I've got are duff and maybe see what it take to break one (static carelessness wise) to see if thats likely to have happened to the ones in there.
Actually something has just occured to me. Can tv coax be leaky if its cheap? Thats something i havent tried unplugging from the wall, and that does run behind all the stereo equipment, probably getting jostled about everytime i remove the thing to take to the test bench.
I think its unlikely as the problem follows the preamp channel rather than the cables.
I have certainly thought this fizzing sound has a slightly digital sounding nature to it which is why i removed my mains networking sockets in an attempt to clear it up (it didnt, but i no longer use them anyway).
Cheers
ECC88 Grid Voltage
To you guys who successfully built the Impasse, I'm curious if you measured the grid voltage of the ECC88 tube? How did it compare to the 118 VDC that SY talks about in the article?
I ask because I have been investigating using a little lower current on the 6SN7. I have a 5692 tube for the 6SN7 position that is more comfortable at a little lower current. Anyway, the lower current CCS would result in a slightly lower voltage on the 6SN7 plate and that would affect the voltage divider R8/R9 which controls the ECC88 grid voltage.
When I went to look at how to adjust the R8/R9 voltage divider, I calculated the original DC voltage at the middle of the voltage divider and the input of the grid. Using the 180 VDC mentioned in the article along with the 680k and 1M for R8 and R9, I get a grid voltage of no more than 107 VDC. I'm sure I'm doing something wrong. What did you guys find?
Jac
To you guys who successfully built the Impasse, I'm curious if you measured the grid voltage of the ECC88 tube? How did it compare to the 118 VDC that SY talks about in the article?
I ask because I have been investigating using a little lower current on the 6SN7. I have a 5692 tube for the 6SN7 position that is more comfortable at a little lower current. Anyway, the lower current CCS would result in a slightly lower voltage on the 6SN7 plate and that would affect the voltage divider R8/R9 which controls the ECC88 grid voltage.
When I went to look at how to adjust the R8/R9 voltage divider, I calculated the original DC voltage at the middle of the voltage divider and the input of the grid. Using the 180 VDC mentioned in the article along with the 680k and 1M for R8 and R9, I get a grid voltage of no more than 107 VDC. I'm sure I'm doing something wrong. What did you guys find?
Jac
lehmanhill, I will probably have mine back on the bench in a few days, ill be sure to have a look.
Ive been needing to make a psu for a turntable and have been reading lots about CL60s and the like. I noticed i had put one inbetween the primaries on my allied transformer for this amp like i did with my F4.
Im wondering if that was a bad idea. As the current draw on the primaries is probably quite low, im wondering if the CL60 gets hot enough for the resistance to drop.
Is it even needed as the psu only has a few hundred uF?
Ive also discovered that putting caps across switches is a bad thing too. I think i currently have an X2 0.01uF 275~ wima there.
What should i have done for this?
Ive been needing to make a psu for a turntable and have been reading lots about CL60s and the like. I noticed i had put one inbetween the primaries on my allied transformer for this amp like i did with my F4.
Im wondering if that was a bad idea. As the current draw on the primaries is probably quite low, im wondering if the CL60 gets hot enough for the resistance to drop.
Is it even needed as the psu only has a few hundred uF?
Ive also discovered that putting caps across switches is a bad thing too. I think i currently have an X2 0.01uF 275~ wima there.
What should i have done for this?
Having had a read through
http://www.diyaudio.com/forums/power-supplies/204603-arcing-mains-switch-add-cap-parallel-why.html
There are a lot of comments about the caps leaking current and essentially stopping the switch from properly isolating.
Another comment on
http://www.diyaudio.com/forums/diyaudio-store/247280-power-supply-soft-start-board-v3-20.html
post #200 also suggests a similar thing.
Apparently I should be using a dual pole mains switch too because as it stands neutral is still connected when i "power off". I think its a requirement in the uk.
Once ive decided what to do about that cap ill sort the switch out.
http://www.diyaudio.com/forums/power-supplies/204603-arcing-mains-switch-add-cap-parallel-why.html
There are a lot of comments about the caps leaking current and essentially stopping the switch from properly isolating.
Another comment on
http://www.diyaudio.com/forums/diyaudio-store/247280-power-supply-soft-start-board-v3-20.html
post #200 also suggests a similar thing.
Apparently I should be using a dual pole mains switch too because as it stands neutral is still connected when i "power off". I think its a requirement in the uk.
Once ive decided what to do about that cap ill sort the switch out.
Thanks Dave. I think I resolved this question for myself. I drew a load line including limits for the 7DJ8 tube I am planning to use. It becomes pretty clear that I need to have an operating point at about -2 V on the grid in order to maintain the proper output voltage swing. With the cathode at 120 V, that means I need to have the grid at about 118 V as stated in the article. I will adjust the voltage divider as needed to achieve this.
What confuses me is that using 180 V at the plate of the 6SN7 and the 680k/1M voltage divider, I calculate 107V at the 6DJ8 grid. That would be almost -13 V which would be more than ideal on the load line I drew. Clearly, I am missing something. But that's why they make sockets, so you can try different values to get the right operating point.
Jac
Last edited:
