lgreen said:
As I indicated earlier, the voltage of the regulated supply is set by the zener diodes. I see no reason for this voltage to vary with time after the initial warm up period.
OK, will hook up the scope to the rats nest in the next step. next you are going to ask me to hook up the signal generator to the input and a resistor to the output? ok ok ok.
I have already tested with the input shorted to ground. No difference in DC offset.
I have already tested with the input shorted to ground. No difference in DC offset.
AndrewT -- from the posts I've read, you're usually right
I short first just to get rid of that variable. If all ok, then active load as you and Steve do. I'm a little cautious here since my system is dc-coupled (with servos) from MC input to speaker.
I asked lgreen to try it because when I had offset problems, I saw a difference in active source vs. grounded input.
Hi
I had an offset problem too. My preamplifier has a low value resistor to ground (1K) on the output and when connected to the Krill input i could set offset to 0v, however when disconnected, offset raised a lot (about 1V) so i used an input 3,3 uF capacitor, adjusted offset and it still is like that. I can connect or disconnect the preamplifier without offset change.
Input resistance to ground has a high influence on offset.
I had an offset problem too. My preamplifier has a low value resistor to ground (1K) on the output and when connected to the Krill input i could set offset to 0v, however when disconnected, offset raised a lot (about 1V) so i used an input 3,3 uF capacitor, adjusted offset and it still is like that. I can connect or disconnect the preamplifier without offset change.
Input resistance to ground has a high influence on offset.
That's almost exactly what I observed, including the effect of the capacitor.
lgreen -- you might try measuring the voltage at the junction of the 56V zener and base of pass transistor while also measuring the voltage at the emitter of the same pass transistor. As Steve said, the zener voltage should be stable after warm-up. If the emitter voltage is drifting around, then there is a problem "somewhere". Try on both power supplies. You could even put that 3rd DMM on the collector of the pass transistor and observe all 3 voltages simultaneously.
I hope you don't mind this question but have you put all 3 DMMs on the same voltage to see if they all measure the same?
I looked over carefully the hi-res pic you linked to and nothing raised any flags.
Phil
Is that a 3.3 uF at the input or is it from input to ground? Must be to ground if its affecting the offset by itself.
OK I'll try that fun little idea tonight.
About the DMMs, I don't mind at all, you guys are the smart ones and I do stupid stuff all the time. I was thinking the same thing a few days ago and have been swapping them up for the different readings- all come out about the same. One is autoranging and takes longer to give a readout but its accurate. I've got a Tek lab DMM in with the FG so if I hook up the FG I can have 4 DMMs... if I really need to go 3 decimal places...
You know what they say- if you have 1 clock you know what time it is, if you have more you never know the time.
OK I'll try that fun little idea tonight.
About the DMMs, I don't mind at all, you guys are the smart ones and I do stupid stuff all the time. I was thinking the same thing a few days ago and have been swapping them up for the different readings- all come out about the same. One is autoranging and takes longer to give a readout but its accurate. I've got a Tek lab DMM in with the FG so if I hook up the FG I can have 4 DMMs... if I really need to go 3 decimal places...
You know what they say- if you have 1 clock you know what time it is, if you have more you never know the time.
lgreen said:
The cap should be at the input, not to ground.
The offset changes because you go from a low impedance (at DC) to a much higher impedance with the cap. This is why I suggest using your preamp as the input while setting the offset. If your preamp is cap coupled, then the added cap should have no effect. If your preamp is DC coupled (as mine and Phil's are), the offset will be different with or without the cap. If the offset can't be adjusted with the input shorted, there is indeed a problem somewhere.
More Tests
Steve probably wishes I never started this!
Here are some tests with very large pics so everyone can read the settings.
- Starting with the loop DMM on the left, that has the bias voltage;
- the green/black one next to it has the DC offset;
- the one above it has the + DC voltage on the primary positive power supply (50V typical at full power).
Here is Vout with 1kHz input signal from iPod at 5 V/Div no load (open circuit on the output). The volume on the iPod is cranked up and the Variac is up to 120VAC
Image 1
Here is Vout with 1Khz input signal from iPod at 5 V/Div with 15 ohm load on the output. Here the input signal is very very low (iPod volume turned way down) and the variac is only turned up enough to get you 10VDC from the primary power supply.
Image 2
Ah yes, drink this one in. ugh.
Oh, and even in the first instance, which looks very nice, you cannot adjust DC offset below about .235 volts and bias is limited to about .7 volts.
But that aside, here is Vout with the input open circ and the output open circ. 50 mV/ Div. Full AC power.
Image 3
So its not that noisy but enough to conclude that something is amiss? Perhaps oscillating?
And here is Vout with the input grounded via a 10K resistor and the output open circ.
Image 4
Noise is less here we are at 20 mV/div instead of 50 in the prior pic. Full AC power.
And here is Vout with the input grounded via a 10K resistor and the ouput connected to 15 ohms. Again, this happened as the variac was just turning up, DC is <20 V out of the 50 it is supposed to be.
Image 5
Ok, now lets talk about the power supplies. I measured the BCE of the pass transistors for the + and - supplies.
For each as the collector goes from 90 to 70 (via the variac) the regulated voltage on the emitters drops a small amount. The pos supply drops from 54.2 to 53.8 and the negative supply drops from -55.4 to -55.0. The emitter drop is about .1 for every 5 volts that drops on the collector.
I have the BCE data at a number of points should you wish to see it.
The supply rails, by the way, look ok in terms of noise and oscillations. A small amount of ripple at the most sensitive scope settings.
I tried tracing back the circ with no input and the 15 ohm load to locate where the oscillations were coming from.
- They were present at R6 and R7 where they feed into Q1 and Q2.
- They were present at the junction of R3 and R4.
- They were present at R11.
- They were not present after R1 (whew that would have been bad).
I have pics of thes if you would like to see them.
Hey, what about the issue with DC/AC coupling of R10, my R10 goes to a 220 uF cap/ to ground, any prob with that?
Steve probably wishes I never started this!
Here are some tests with very large pics so everyone can read the settings.
- Starting with the loop DMM on the left, that has the bias voltage;
- the green/black one next to it has the DC offset;
- the one above it has the + DC voltage on the primary positive power supply (50V typical at full power).
Here is Vout with 1kHz input signal from iPod at 5 V/Div no load (open circuit on the output). The volume on the iPod is cranked up and the Variac is up to 120VAC
Image 1
Here is Vout with 1Khz input signal from iPod at 5 V/Div with 15 ohm load on the output. Here the input signal is very very low (iPod volume turned way down) and the variac is only turned up enough to get you 10VDC from the primary power supply.
Image 2
Ah yes, drink this one in. ugh.
Oh, and even in the first instance, which looks very nice, you cannot adjust DC offset below about .235 volts and bias is limited to about .7 volts.
But that aside, here is Vout with the input open circ and the output open circ. 50 mV/ Div. Full AC power.
Image 3
So its not that noisy but enough to conclude that something is amiss? Perhaps oscillating?
And here is Vout with the input grounded via a 10K resistor and the output open circ.
Image 4
Noise is less here we are at 20 mV/div instead of 50 in the prior pic. Full AC power.
And here is Vout with the input grounded via a 10K resistor and the ouput connected to 15 ohms. Again, this happened as the variac was just turning up, DC is <20 V out of the 50 it is supposed to be.
Image 5
Ok, now lets talk about the power supplies. I measured the BCE of the pass transistors for the + and - supplies.
For each as the collector goes from 90 to 70 (via the variac) the regulated voltage on the emitters drops a small amount. The pos supply drops from 54.2 to 53.8 and the negative supply drops from -55.4 to -55.0. The emitter drop is about .1 for every 5 volts that drops on the collector.
I have the BCE data at a number of points should you wish to see it.
The supply rails, by the way, look ok in terms of noise and oscillations. A small amount of ripple at the most sensitive scope settings.
I tried tracing back the circ with no input and the 15 ohm load to locate where the oscillations were coming from.
- They were present at R6 and R7 where they feed into Q1 and Q2.
- They were present at the junction of R3 and R4.
- They were present at R11.
- They were not present after R1 (whew that would have been bad).
I have pics of thes if you would like to see them.
Hey, what about the issue with DC/AC coupling of R10, my R10 goes to a 220 uF cap/ to ground, any prob with that?
I`ve done listening tests of krill for a two months now and then tried to listen "classic" diamond based triple emiter-follower. Triple sounds better with more space and more details. That`s my subjective opinion.
More on the PS
I was thinking that the regulated PS might be collapsing when a load is connected.
So I loaded down the regulated rails.
for pos rail
R=9.5K, V drops from 54.0 to 53.9
R=2.15K, V drops from 54.0 to 53.8
R =1K, V drops from 54.0 to 53.6
R = 664, V drops from 54.0 to 51.3
for neg rail
R=9.5K, V drops -55.6 to -55.5
R=2.15K, V drops -55.6 to -55.3
R=1K, V drops -55.6 to -54.9
R=664, V drops -55.6 to -50.0
I was thinking that the regulated PS might be collapsing when a load is connected.
So I loaded down the regulated rails.
for pos rail
R=9.5K, V drops from 54.0 to 53.9
R=2.15K, V drops from 54.0 to 53.8
R =1K, V drops from 54.0 to 53.6
R = 664, V drops from 54.0 to 51.3
for neg rail
R=9.5K, V drops -55.6 to -55.5
R=2.15K, V drops -55.6 to -55.3
R=1K, V drops -55.6 to -54.9
R=664, V drops -55.6 to -50.0
There should be no problem with the cap to ground. It should be a good high quality of cap.
Try adding the compensation cap, C2 I believe. Start with something near 30pF or greater.
Try adding the compensation cap, C2 I believe. Start with something near 30pF or greater.
C2
I tried putting caps in for C2 (had to take the heat sink off the nearby transistor). I had a few small value caps around and tried them.
100 pf - nothing
220 pf- nothing/ oscillations changed to pure sine waves of high freq.
330 pf- somewhat smaller osciallations where PS is <10V but did nothing as the PS increased from 10 to 50V
3300 pf (.0033uF) - all gone, yes!
I used a 50V polyproplene cap -
Digikey P3907-ND
Good or start buying smaller or more HQ caps to see if they work?
50V rating ok?
I still have the problems with bias and offset though.
tried caps on the input but they did nothing.
So we are back to the original problem, bias and offset. The amp had no problem driving 40V pk-pk into 15 ohms (1KHz).
I tried putting caps in for C2 (had to take the heat sink off the nearby transistor). I had a few small value caps around and tried them.
100 pf - nothing
220 pf- nothing/ oscillations changed to pure sine waves of high freq.
330 pf- somewhat smaller osciallations where PS is <10V but did nothing as the PS increased from 10 to 50V
3300 pf (.0033uF) - all gone, yes!
I used a 50V polyproplene cap -
Digikey P3907-ND
Good or start buying smaller or more HQ caps to see if they work?
50V rating ok?
I still have the problems with bias and offset though.
tried caps on the input but they did nothing.
So we are back to the original problem, bias and offset. The amp had no problem driving 40V pk-pk into 15 ohms (1KHz).
lgreen --
I'm thinking that most if not all of the problems you observe are related to the inability to set the output idle current. I saw much of the same behavior you do and most of it went away when I got the output dialed in.
Try measuring the offset voltage at the "output end" of R11, i.e., the junction of R11, Q4, Q6, Q7 and Q10. Check its adjustment range and see if it is different than the offset measured at the amp's offset.
It's really helpful to be able to separate the voltage amplifier from the output stage. I cut that long straight trace to the junction of Q7 and Q10 and put a resistor from the junction of Q7 and Q10 to signal common then used the junction of the resistor, Q7 and Q10 as the input to the output stage. The resistor value isn't too critical -- I used 100K. This might be more surgery than you want to do but reducing the number of variables in the problem is the shortest path to a solution.
After you do this, I suspect the voltage amplifier will work fine with no or a very small compensation capacitor and that you need some tweaking on the output stage to get the idle current set right. It's not a matter of changing any circuit values but finding the right thermal conditions on the TO-220s and the diode string. It's a case where the schematic doesn't tell the whole story.
I'm thinking that most if not all of the problems you observe are related to the inability to set the output idle current. I saw much of the same behavior you do and most of it went away when I got the output dialed in.
Try measuring the offset voltage at the "output end" of R11, i.e., the junction of R11, Q4, Q6, Q7 and Q10. Check its adjustment range and see if it is different than the offset measured at the amp's offset.
It's really helpful to be able to separate the voltage amplifier from the output stage. I cut that long straight trace to the junction of Q7 and Q10 and put a resistor from the junction of Q7 and Q10 to signal common then used the junction of the resistor, Q7 and Q10 as the input to the output stage. The resistor value isn't too critical -- I used 100K. This might be more surgery than you want to do but reducing the number of variables in the problem is the shortest path to a solution.
After you do this, I suspect the voltage amplifier will work fine with no or a very small compensation capacitor and that you need some tweaking on the output stage to get the idle current set right. It's not a matter of changing any circuit values but finding the right thermal conditions on the TO-220s and the diode string. It's a case where the schematic doesn't tell the whole story.
one more thought along the lines of reducing variables........
Are you using the power supply in your post 281? If so, try connecting one end of each secondary together -- paying attention to getting the phases right -- and connecting that to the common point on the board. You can keep the two FWBs and other connections as in post 281.
Try this before the more drastic approach in my previous post.
Are you using the power supply in your post 281? If so, try connecting one end of each secondary together -- paying attention to getting the phases right -- and connecting that to the common point on the board. You can keep the two FWBs and other connections as in post 281.
Try this before the more drastic approach in my previous post.
PH104 said:
Yes, I measured this in Post 332.
----> Measuring R11 at the top as suggested. The DC value ranges from 7.4 mV to 90 mV as I turned the offset pot clockwise. It did not go through zero or get negative, just goes up/down.
The offset there is much lower.
Also, I am NOT using the power supply posted in #281. Instead I took the other bridge out and am using just one bridge. The AC from the transformer goes into the AC input on the board, the rectified AC from the single bridge goes into the large capacitors (the output end of where the on-board diodes would have been). Seems to work just fine.
Re: Re: C2
The quality is OK, but the voltage should be at least twice the voltage of your regulated supply. I think that after you get the bias to set, that cap will not need to be over about 30pf.
lgreen said:
The quality is OK, but the voltage should be at least twice the voltage of your regulated supply. I think that after you get the bias to set, that cap will not need to be over about 30pf.