The capacitor is used to "losslessly" reduce the voltage across the reactance of the capacitor, since reactance exists in the imaginary plane not the real plane, there is no power dissipated in the reactance.
HOWEVER and this is a big one, there is always some ESR in capacitors and the power dissipated across this is real, and the capacitor may warm significantly depending on the amount of esr present.
No single capacitor of any brand I tried lived for long in this circuit which is why I had several of them in series parallel to do this. Note that I added the resistor at the same time as I increased the number of capacitors. Circuits not using the clamp diode may not need that resistor, but I got much more consistent behavior with the topology shown than with a simple resistor from the cap to ground.
I used this technique because the reactance allowed me to control the rectified voltage fairly effectively without using big resistors, very high voltage diodes and high voltage capacitors. (Take a look at the circuit in my amplifier and you will see what I mean - note however that the schematic does not show the 470 ohm resistor I put in series with the capacitor.) One downside of this technique is that it is somewhat sensitive to line frequency, in my design I would recommend increasing the capacitance to 0.12uF if operation is at 50Hz.
Kevin
HOWEVER and this is a big one, there is always some ESR in capacitors and the power dissipated across this is real, and the capacitor may warm significantly depending on the amount of esr present.
No single capacitor of any brand I tried lived for long in this circuit which is why I had several of them in series parallel to do this. Note that I added the resistor at the same time as I increased the number of capacitors. Circuits not using the clamp diode may not need that resistor, but I got much more consistent behavior with the topology shown than with a simple resistor from the cap to ground.
I used this technique because the reactance allowed me to control the rectified voltage fairly effectively without using big resistors, very high voltage diodes and high voltage capacitors. (Take a look at the circuit in my amplifier and you will see what I mean - note however that the schematic does not show the 470 ohm resistor I put in series with the capacitor.) One downside of this technique is that it is somewhat sensitive to line frequency, in my design I would recommend increasing the capacitance to 0.12uF if operation is at 50Hz.
Kevin
poobah,
When first built I had Sprague OD, .05uf 600v.
The bias was always a bit funky. The under-rated diodes didn't help much either...
When I would take a reading at the test points for plate voltage I always got a rock steady reading, but when I'd check test points for the voltage drop on the 27k cathode resistors to ground I'd get voltage that would fluctuate. If I had the DC balance set to .905 for each 7591, it would constantly fluctuate up and down to .910-.899-.905-.915 and so on.
The bias continued to move more negatively all the time as well, causing me to constanly monitor and adjust-I knew this wasn't right but I also knew I followed the schematic exactly so I knew it wasn't a mistake in the build.
A couple weeks ago I was checking and adjusting bias and one channel stopped and I shut it down. Went checking and found the .05 600v orange drops had shorted.
Replaced them with a pair of .022uf 600v in parallel and brought the amp up with my variable AC power supply and checked voltages-everything was in line so I listened for a few minutes, then checked the plate test points and read 560v when I should read 480. Shut it down, checked the new caps-no shorts-and started suspecting the diodes.
On Saturday I replaced the 1,000piv diodes with 1,500piv and brought things up slowly on my variable AC power supply. After a few tests things looked good so I gave it a listen, but found the test points for the 7591 cathode resistors are still fluctuating as before.
When first built I had Sprague OD, .05uf 600v.
The bias was always a bit funky. The under-rated diodes didn't help much either...
When I would take a reading at the test points for plate voltage I always got a rock steady reading, but when I'd check test points for the voltage drop on the 27k cathode resistors to ground I'd get voltage that would fluctuate. If I had the DC balance set to .905 for each 7591, it would constantly fluctuate up and down to .910-.899-.905-.915 and so on.
The bias continued to move more negatively all the time as well, causing me to constanly monitor and adjust-I knew this wasn't right but I also knew I followed the schematic exactly so I knew it wasn't a mistake in the build.
A couple weeks ago I was checking and adjusting bias and one channel stopped and I shut it down. Went checking and found the .05 600v orange drops had shorted.
Replaced them with a pair of .022uf 600v in parallel and brought the amp up with my variable AC power supply and checked voltages-everything was in line so I listened for a few minutes, then checked the plate test points and read 560v when I should read 480. Shut it down, checked the new caps-no shorts-and started suspecting the diodes.
On Saturday I replaced the 1,000piv diodes with 1,500piv and brought things up slowly on my variable AC power supply. After a few tests things looked good so I gave it a listen, but found the test points for the 7591 cathode resistors are still fluctuating as before.
Well Mojo,
I ask about the Orange drops 'cause there are many types that go under the same name. And after hearing the horror stories from Kevinkr, I am reluctant to tell you to throw a bigger (voltage)cap in... Although a cap with a good AC rating should do it.
Do you still have your old caps and can you give us the number on them? Did you happen to touch them while running... if you do so... do it carefully!!!!
We can calculate the whole nine yards; which seems a smart thing at this point.
1. Do you know what voltage you're trying to get out of the circuit?
2. Do you know the current that is flowing?
I ask about the Orange drops 'cause there are many types that go under the same name. And after hearing the horror stories from Kevinkr, I am reluctant to tell you to throw a bigger (voltage)cap in... Although a cap with a good AC rating should do it.
Do you still have your old caps and can you give us the number on them? Did you happen to touch them while running... if you do so... do it carefully!!!!
We can calculate the whole nine yards; which seems a smart thing at this point.
1. Do you know what voltage you're trying to get out of the circuit?
2. Do you know the current that is flowing?
kevin,
You're experience with this type of ciruit is invaluable. Would 3 of the 1600VDC/600VAC .015 Orange Drops in parallel do the trick or would I have to go with 5 .01s to be totally safe?
It's heartening to know that while this isn't neccessarily the most efficient way of developing bias it CAN be a viable solution.
This was my first DIY amp and I built it totally on my own-I had some help from members here with a couple other trouble shooting issues once it was built, but other than that I didn't have anybody to guide me. I had Heyboer wind the transformers for me, and other than over-rating for current, I had them make an exact copy of the power transformer listed in the BOM.
Didn't know any better then, but now I would've had a 55v bias secondary added to the power transformer or I would've incorporated a separate bias transformer. But since the amp wasn't laid out around those considerations I'd like to make it work with the layout as is if possible since I spent so much time planning out the build to make things as neat as possible.
Best,
mr mojo
You're experience with this type of ciruit is invaluable. Would 3 of the 1600VDC/600VAC .015 Orange Drops in parallel do the trick or would I have to go with 5 .01s to be totally safe?
It's heartening to know that while this isn't neccessarily the most efficient way of developing bias it CAN be a viable solution.
This was my first DIY amp and I built it totally on my own-I had some help from members here with a couple other trouble shooting issues once it was built, but other than that I didn't have anybody to guide me. I had Heyboer wind the transformers for me, and other than over-rating for current, I had them make an exact copy of the power transformer listed in the BOM.
Didn't know any better then, but now I would've had a 55v bias secondary added to the power transformer or I would've incorporated a separate bias transformer. But since the amp wasn't laid out around those considerations I'd like to make it work with the layout as is if possible since I spent so much time planning out the build to make things as neat as possible.
Best,
mr mojo
poobah,
I sure appreciate the help! IIRC correctly, I've got -55v at the bias adjustment pots and the schematic lists 45ma ratings for the diodes. The 7591s bias @-18 to -20v for 33ma per tube.
I'll open it up tonight and check the voltage drop across the resistors going to the grids and get a more accurate idea of the bias current.
Best,
mr mojo
I sure appreciate the help! IIRC correctly, I've got -55v at the bias adjustment pots and the schematic lists 45ma ratings for the diodes. The 7591s bias @-18 to -20v for 33ma per tube.
I'll open it up tonight and check the voltage drop across the resistors going to the grids and get a more accurate idea of the bias current.
Best,
mr mojo
Tweeker,
I'm totally open to using something better suited-just initially thought about the 1600v OD since I can order them from Antique Electronics. If I went with the Cornell Dubilier 942Cs would I need to run them in parallel as kevin suggested or do you think the .047s I saw listed do the trick?
I sure appreciate your input as well-hope you didn't think I was ignoring your post-just trying to keep up with the others and trying to do something that at least looks like work here at work keeps me a few posts behind sometimes-HA!
Best,
mr mojo
I'm totally open to using something better suited-just initially thought about the 1600v OD since I can order them from Antique Electronics. If I went with the Cornell Dubilier 942Cs would I need to run them in parallel as kevin suggested or do you think the .047s I saw listed do the trick?
I sure appreciate your input as well-hope you didn't think I was ignoring your post-just trying to keep up with the others and trying to do something that at least looks like work here at work keeps me a few posts behind sometimes-HA!
Best,
mr mojo
The Cornell Dubilier caps would work, but are kind of hard to find in that value in small quantity. Mouser has .1uF 2,000VDC 942Cs, but its too expensive to series them to get .05uF.
Digikey has has Panasonic ECW-H16473JV in stock, which is what I'd substitute. 1600VDC, .047uF. These capacitors are suited for electronic ballast use. In testing the slightly larger .056uF 1600V Panasonic capacitor 3.5 amps RMS of 350khz sine wave continous lead to a 3.1 degree C temperature rise. $2.78 each. Pushed, they can take 6.3I RMS, short term. This is in real testing, not simulation.
Digikey has has Panasonic ECW-H16473JV in stock, which is what I'd substitute. 1600VDC, .047uF. These capacitors are suited for electronic ballast use. In testing the slightly larger .056uF 1600V Panasonic capacitor 3.5 amps RMS of 350khz sine wave continous lead to a 3.1 degree C temperature rise. $2.78 each. Pushed, they can take 6.3I RMS, short term. This is in real testing, not simulation.
Hey Mojo,
I ran some numbers... assuming I read the resistor values right.
The total current flowing out of this circuit to the bias network is about 590 uA. So for the most part the caps see the first 27K resistor, with their 5 mA flow. Its just a simple AC thing at this point.
The caps have an odd waveform 270 VAC across them on one phase and a MAX of 380 VAC on the other...
So, if you had 225P orange drops they would have died.
If you had 715P's Orange drops with a DC rating of 600 Volts, and an AC rating of 200 Volts they would have died...especially when the inside of the amp heated up a bit.
Now if you use 715P's, 1200 VDC, 475 AC (rms), you should be good to go...
p/n 715P1200V0.047
Mouser $2.02
Now these voltage ratings only hold true up to 85 deg C. but if your amp was that hot you'ld be cooking Christmas roast or something.
You could put 2 - 0.022 uF's with 1600 VDC but it won't buy you much on the VAC rating... 500 volts instead of 475.
These caps are big though... 1.7" x .9" and that's the key to the heat problem... more surface area.
Cheers,
I ran some numbers... assuming I read the resistor values right.
The total current flowing out of this circuit to the bias network is about 590 uA. So for the most part the caps see the first 27K resistor, with their 5 mA flow. Its just a simple AC thing at this point.
The caps have an odd waveform 270 VAC across them on one phase and a MAX of 380 VAC on the other...
So, if you had 225P orange drops they would have died.
If you had 715P's Orange drops with a DC rating of 600 Volts, and an AC rating of 200 Volts they would have died...especially when the inside of the amp heated up a bit.
Now if you use 715P's, 1200 VDC, 475 AC (rms), you should be good to go...
p/n 715P1200V0.047
Mouser $2.02
Now these voltage ratings only hold true up to 85 deg C. but if your amp was that hot you'ld be cooking Christmas roast or something.
You could put 2 - 0.022 uF's with 1600 VDC but it won't buy you much on the VAC rating... 500 volts instead of 475.
These caps are big though... 1.7" x .9" and that's the key to the heat problem... more surface area.
Cheers,
use small transformers
a very small transformer, say 6.3volts sec. and 110volts pri connected backwards, powered from the 6.3 volt heater supply and the 110volt half wave rectified will do the trick, an 1n4936 is suitable, then crc filtering and trimpots and you are in bussines!!
no more capacitor woes!
a very small transformer, say 6.3volts sec. and 110volts pri connected backwards, powered from the 6.3 volt heater supply and the 110volt half wave rectified will do the trick, an 1n4936 is suitable, then crc filtering and trimpots and you are in bussines!!
no more capacitor woes!
Tony and Tom,
The transformer idea is a great one-only issue is space. As you can see in the pic, the amp was laid out around the capacitor solution so I'd like to make it work if at all possible.
In the meantime, I've ordered some 3000VDC/750VAC .047uf Cornell Dubilier caps from Allied-should be here tomorrow. I also ordered six .015 1,600VDC/700VAC Sprague Orange drops to run 3 in parallel if the Cornells don't have enough surface area to dissapate the heat.
I really appreciate everyone's help on this-it's just amazing to be able to come here, ask a question and get so many friendly, knowledgeable and helpful responses. You folks are great and you have my most sincere thanks!
Hopefully I'll update this thread on Wednesday and let you know how things worked out.
Best,
mr mojo
The transformer idea is a great one-only issue is space. As you can see in the pic, the amp was laid out around the capacitor solution so I'd like to make it work if at all possible.
An externally hosted image should be here but it was not working when we last tested it.
In the meantime, I've ordered some 3000VDC/750VAC .047uf Cornell Dubilier caps from Allied-should be here tomorrow. I also ordered six .015 1,600VDC/700VAC Sprague Orange drops to run 3 in parallel if the Cornells don't have enough surface area to dissapate the heat.
I really appreciate everyone's help on this-it's just amazing to be able to come here, ask a question and get so many friendly, knowledgeable and helpful responses. You folks are great and you have my most sincere thanks!
Hopefully I'll update this thread on Wednesday and let you know how things worked out.
Best,
mr mojo
tweeker,
Thanks for the pic-it's gonna take a little creative manuevering to get these bigger caps in there...
One other thing I've been thinking about; and maybe this is what you guys have been trying to get through to me all this time-like I said, I'm no engineer!
Dissapation factor. Schematic lists PIO caps since in 1961 that's all that was available. But PIO also has a lower dissapation factor than poly/film caps: .05% for PIO and .1% for poly. Maybe that's what enables a 600v PIO cap to survive.
So I got to thinking, maybe the best route is three .015uf 1600VDC/700VAC poly caps in parallel-as Kevin suggested. That should give me a dissapation factor of .033% and at a suitable VAC rating as well.
What're your thoughts?
Best,
mr mojo
Thanks for the pic-it's gonna take a little creative manuevering to get these bigger caps in there...
One other thing I've been thinking about; and maybe this is what you guys have been trying to get through to me all this time-like I said, I'm no engineer!
Dissapation factor. Schematic lists PIO caps since in 1961 that's all that was available. But PIO also has a lower dissapation factor than poly/film caps: .05% for PIO and .1% for poly. Maybe that's what enables a 600v PIO cap to survive.
So I got to thinking, maybe the best route is three .015uf 1600VDC/700VAC poly caps in parallel-as Kevin suggested. That should give me a dissapation factor of .033% and at a suitable VAC rating as well.
What're your thoughts?
Best,
mr mojo
Foil vs. metalized is the key here... along with obvious issue of AC ratings (which we have all thoroughly beat to death)... you're good to go here Mojo... relax. Put in the 2 cornies and away you go!
Put 'em in and run it... if they get warm, that's FINE. if they get hot (they won't), go nuts with the parallel stuff. Heat is what consumes voltage and current ratings in caps. 'Sides it would suck to screw up your pretty amp with a kluge of caps hanging all over the place.
Oh, and you still (might) need to replace those pots... 'cause they got the hell beat out of them along with 27K's that smoked... something has to explain the "erratic" behavior you talked about.
Happy new year guys!
Put 'em in and run it... if they get warm, that's FINE. if they get hot (they won't), go nuts with the parallel stuff. Heat is what consumes voltage and current ratings in caps. 'Sides it would suck to screw up your pretty amp with a kluge of caps hanging all over the place.
Oh, and you still (might) need to replace those pots... 'cause they got the hell beat out of them along with 27K's that smoked... something has to explain the "erratic" behavior you talked about.
Happy new year guys!
Dissipation factor in this application is not a problem once your at polypropylene or better. Some specialized (pulse duty) metallized capacitors are just fine here.
Here are the results of some dissipation testing done on polypropylene pulse capacitors by Terry Fritz:
0.047uF 1500V Arcotronics capacitors "RS" from Richie Burnett
2.5 amps RMS 350kHz 2.2 degrees C rise
0.022uF 1200V Panasonic Q-E?? from Bill Noble
2.0 amps RMS 350kHz 4.8 degrees C rise
0.056uF 1600V Phillips 376 KP/MMKP HQ from Finn Hammer
3.5 amps RMS 350kHz 1.5 degrees C rise
0.056uF 1600V Panasonic ECW-H16563JV stocked
3.5amps RMS 350kHz 3.1 degrees C rise
This is way higher frequency than your application. RMS current here is in series with the capacitor. Do not confuse RMS current with average current.
Your main issue is I2R related, PIO caps are of foil construction so have more conductor than some.
Here are the results of some dissipation testing done on polypropylene pulse capacitors by Terry Fritz:
0.047uF 1500V Arcotronics capacitors "RS" from Richie Burnett
2.5 amps RMS 350kHz 2.2 degrees C rise
0.022uF 1200V Panasonic Q-E?? from Bill Noble
2.0 amps RMS 350kHz 4.8 degrees C rise
0.056uF 1600V Phillips 376 KP/MMKP HQ from Finn Hammer
3.5 amps RMS 350kHz 1.5 degrees C rise
0.056uF 1600V Panasonic ECW-H16563JV stocked
3.5amps RMS 350kHz 3.1 degrees C rise
This is way higher frequency than your application. RMS current here is in series with the capacitor. Do not confuse RMS current with average current.
Your main issue is I2R related, PIO caps are of foil construction so have more conductor than some.
Hi Mojo,
Incidentally the value of that capacitance is quite important, you should stay quite close to the recommended capacitance value or see the voltage increase/decrease dramatically.. If you are using 0.015uF then three in parallel will get you quite close to the original value.
As others here have indicated the construction of the capacitor is important, foil types will handle the current much more readily than any metalized film types.
Incidentally what is important here are the voltage & current waveforms across the capacitor, not the voltage to ground. If you have a two channel scope you can check the voltage waveform across the capacitor using differential input mode with channel a on one side of cap and channel b on other. (On Tek scopes invert channel b and select a+b mode, set both channels to same range.) Make sure you have high voltage probes, connect probe grounds to chassis. To measure current just stick a 10 ohm resistor in series with that cap and connect the probes across it. Pk voltage and current ratings of the capacitor should not be exceeded if known.
Hope this helps..
Incidentally the value of that capacitance is quite important, you should stay quite close to the recommended capacitance value or see the voltage increase/decrease dramatically..
If you are using 0.015uF then three in parallel will get you quite close to the original value.As others here have indicated the construction of the capacitor is important, foil types will handle the current much more readily than any metalized film types.
Incidentally what is important here are the voltage & current waveforms across the capacitor, not the voltage to ground. If you have a two channel scope you can check the voltage waveform across the capacitor using differential input mode with channel a on one side of cap and channel b on other. (On Tek scopes invert channel b and select a+b mode, set both channels to same range.) Make sure you have high voltage probes, connect probe grounds to chassis. To measure current just stick a 10 ohm resistor in series with that cap and connect the probes across it. Pk voltage and current ratings of the capacitor should not be exceeded if known.
Hope this helps..
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.