Ah, ok. Something with high-ish ESR. In that case, i suppose a small resistor in series with a film cap across the output would work? I don't think I have any electrolytic caps in that capacitance range on hand (lots of large capacitance ones, though -- 100 to 220 uF). But I have lots of small value resistors and a bunch of 1 to 20 uF film caps.
cheers and thanks, Derek
cheers and thanks, Derek
Improvement achieved - but still not rock steady.
I increased the current limit to 300 mA, but saw no difference in performance. Output voltage bounced around as much as 2V pk-pk. But this needed to be done in any event b/c I mistakenly used the quiescent current when calculating my max current needs.
I then tried a 10 uF film cap in series with a 2 ohm resistor across the T-reg output. I noticed no difference, but only let the amp run for 5 minutes.
Lastly, I put two 220 uF 400V caps in series (for effective 110 uF 800V) across the output. For the first 5+ minutes there was no difference. But thereafter the voltage settled down. And at 10+ minutes after power on, the voltage stabilized to within 100mV of 450V.
An observation that I find curious: when the output voltage of the T-reg is most unstable, the output seems insensitive to changes to the 5K trimmer resistor. When it is more stable it is more senstive to changes to the trimmer. I wonder whether this points to a fried trimmer - perhaps I used too much heat when soldering it to the board. I have extras an will test what happens if I replace it.
Things I don't know but plan to test:
(1) whether the stabilization of voltage with the effective 110 uF cap is a function of the cap or of time - as this was only test setup that I ran for more than 10 minutes. I will test the stock board for 10+ minutes to see.
(2) Whether the board is functionlng properly (i.e., whether I screwed up the stuffing). I've ordered some high value high power-rated resistors to test how the regulator behaves when it has a purely resistive load.
(3) as mentioned above, whether the 5K got overheated and is misbehaving.
cheers, Derek
I increased the current limit to 300 mA, but saw no difference in performance. Output voltage bounced around as much as 2V pk-pk. But this needed to be done in any event b/c I mistakenly used the quiescent current when calculating my max current needs.
I then tried a 10 uF film cap in series with a 2 ohm resistor across the T-reg output. I noticed no difference, but only let the amp run for 5 minutes.
Lastly, I put two 220 uF 400V caps in series (for effective 110 uF 800V) across the output. For the first 5+ minutes there was no difference. But thereafter the voltage settled down. And at 10+ minutes after power on, the voltage stabilized to within 100mV of 450V.
An observation that I find curious: when the output voltage of the T-reg is most unstable, the output seems insensitive to changes to the 5K trimmer resistor. When it is more stable it is more senstive to changes to the trimmer. I wonder whether this points to a fried trimmer - perhaps I used too much heat when soldering it to the board. I have extras an will test what happens if I replace it.
Things I don't know but plan to test:
(1) whether the stabilization of voltage with the effective 110 uF cap is a function of the cap or of time - as this was only test setup that I ran for more than 10 minutes. I will test the stock board for 10+ minutes to see.
(2) Whether the board is functionlng properly (i.e., whether I screwed up the stuffing). I've ordered some high value high power-rated resistors to test how the regulator behaves when it has a purely resistive load.
(3) as mentioned above, whether the 5K got overheated and is misbehaving.
cheers, Derek
(2) Whether the board is functioning properly (i.e., whether I screwed up the stuffing). I've ordered some high value high power-rated resistors to test how the regulator behaves when it has a purely resistive load.
Yes please, you should do that. Then we know whether it is the reg or something else.
Jan
Yes please, you should do that. Then we know whether it is the reg or something else.
Jan
Mouser delivery finally arrived. With a purely resistive load the T-reg works perfectly. With a 50K resistor load, output voltage was instantly stable: approx. 470VDC input to regulated 450VDC output.
So the issue is with the parallel feed (parafeed) load. I have verified that the amp, including the 300B output stage that wish to regulate, is operating properly.
Lundahl's specs for the plate choke: 53H with a 160 ohm DCR. Edit: The T-reg feeds two of them, one per 300B output stage.
Any ideas as to what to try?
My only idea is to add to the DCR of the choke by putting a 75R resistor in series with it. I have enough raw DC voltage to make this work. It would still leave me with approx. 8 VDC of compliance for the reg.
many thanks, Derek
So the issue is with the parallel feed (parafeed) load. I have verified that the amp, including the 300B output stage that wish to regulate, is operating properly.
Lundahl's specs for the plate choke: 53H with a 160 ohm DCR. Edit: The T-reg feeds two of them, one per 300B output stage.
Any ideas as to what to try?
My only idea is to add to the DCR of the choke by putting a 75R resistor in series with it. I have enough raw DC voltage to make this work. It would still leave me with approx. 8 VDC of compliance for the reg.
many thanks, Derek
Last edited:
Derek, have you tried to place a largish electrolytic like 47uF at the T-reg output?
I believe that the problem with the large inductive load is the fact that an inductor tends to create a large varying voltage with varying current. If that voltage gets above the T-reg output, it can't 'go' anywhere - a regulator is designed to source current, not to sink it.
Jan
I believe that the problem with the large inductive load is the fact that an inductor tends to create a large varying voltage with varying current. If that voltage gets above the T-reg output, it can't 'go' anywhere - a regulator is designed to source current, not to sink it.
Jan
Thanks Jan.
I did try the equivalent of a 110 uF cap across the output of the reg, and that seemed to help, but did not give me a steady output -- see post #243 above.
So I take it that the reg and the choke are fighting each other - the reg varying current to hold voltage constant while the choke is varying voltage to maintain constant current. This makes sense to me, but I am a bit surprised that this tug-of-war occurs even when the output stage is operating at its quiescent point.
Conceptually, I can't see a way around this problem. But then how is the T-reg able to function properly when feeding a "normal" output stage -- i.e., the primary of a gapped output transformer -- doesn't the gapped OT behave like a choke?
cheers and many thanks, Derek
I did try the equivalent of a 110 uF cap across the output of the reg, and that seemed to help, but did not give me a steady output -- see post #243 above.
So I take it that the reg and the choke are fighting each other - the reg varying current to hold voltage constant while the choke is varying voltage to maintain constant current. This makes sense to me, but I am a bit surprised that this tug-of-war occurs even when the output stage is operating at its quiescent point.
Conceptually, I can't see a way around this problem. But then how is the T-reg able to function properly when feeding a "normal" output stage -- i.e., the primary of a gapped output transformer -- doesn't the gapped OT behave like a choke?
cheers and many thanks, Derek
These are good thoughts, and I don't have an immediate answer. The problem can only arise if at any point on the signal the current reversed, i.e. flows back into the regulator, or at least trying to.
In your case, you have both a choke load and an output transformer, right?
Is it possible to run your amp with the output transformer disconnected?
Your #243 seems to point to something akin to motorboating in a tube amp, which is caused by a particular combination of time constants in the power supply and amp circuits. Hmmm.
Maybe you can send me the amp schematic, confidential of course.
Another thing to try is to add say a 10mA DC load to the regulator, by connecting a 40k or 50k resistor across its output.
That gives the reg a 10mA reverse current capability.
Jan
In your case, you have both a choke load and an output transformer, right?
Is it possible to run your amp with the output transformer disconnected?
Your #243 seems to point to something akin to motorboating in a tube amp, which is caused by a particular combination of time constants in the power supply and amp circuits. Hmmm.
Maybe you can send me the amp schematic, confidential of course.
Another thing to try is to add say a 10mA DC load to the regulator, by connecting a 40k or 50k resistor across its output.
That gives the reg a 10mA reverse current capability.
Jan
Last edited:
Many thanks Jan.
Edit: yes, I have both a plate choke and a zero-gapped (parafeed) output transformer.
I can definitely try a 50K resistor across the output in parallel with the parafeed output stage. And I suspect I can also test with the output transformer disconnected -- but I want to check with some parafeed experts for confirmation that it is ok to do this.
I will post back when I have done one or both of the above tests.
cheers and thanks, Derek
Edit: yes, I have both a plate choke and a zero-gapped (parafeed) output transformer.
I can definitely try a 50K resistor across the output in parallel with the parafeed output stage. And I suspect I can also test with the output transformer disconnected -- but I want to check with some parafeed experts for confirmation that it is ok to do this.
I will post back when I have done one or both of the above tests.
cheers and thanks, Derek
Last edited:
No luck with the tests. All of the following failed: (1) parafeed load (amp as designed) with 50K resistor across T-reg output; (2) same as (1) but with addition of 110 uF cap in parallel with 50K resistor; (3) same as (2) but with output transformers disconnected. In all cases, the T-reg output was unstable - randomly bouncing around +/- 1V rms or so.
I will look into the motorboating issue. I think I will need to scope the amp to explore this. Without the reg, with the output stage fed directly from the CLCLC filter, I don't hear any motorboating, or hum/noise of any kind. But I suppose there could be inaudible oscillation. I need to read up more about this and do some scoping. This will have to wait a week or more -- but I will report back on what I find.
cheers and thanks, Derek
I will look into the motorboating issue. I think I will need to scope the amp to explore this. Without the reg, with the output stage fed directly from the CLCLC filter, I don't hear any motorboating, or hum/noise of any kind. But I suppose there could be inaudible oscillation. I need to read up more about this and do some scoping. This will have to wait a week or more -- but I will report back on what I find.
cheers and thanks, Derek
I think that will only be the case when the DC input voltage rises slower than the ref voltage which is normally not the case. I believe that at startup, the output voltage will always be lower than the input voltage. What are you getting at?
@Derek: can you try to set the T-reg to a much lower voltage, so that is it much lower than its DC input? Like 50V lower than the input? Does that make a difference?
Jan
@Derek: can you try to set the T-reg to a much lower voltage, so that is it much lower than its DC input? Like 50V lower than the input? Does that make a difference?
Jan
Last edited:
In simulation it looked as if it took a brief period of time for the LT3092 to get to the steady state current. Might not be the case in real life with some nanoHenries of connections between B+ and the T-reg.
Datasheet also states that quieting the current source with a small capacitor across Rset adds to startup time.
They were expected in the diyaudio store but will take some time.
I have a number of sets available as described on my website, so first come first served.
Complete with AD8031 in socket, pass device, low current LED and high voltage current source device; LT3092 and reset pushbutton pre-soldered (SMD).
€ 35 + shipping.
Jan
I have a number of sets available as described on my website, so first come first served.
Complete with AD8031 in socket, pass device, low current LED and high voltage current source device; LT3092 and reset pushbutton pre-soldered (SMD).
€ 35 + shipping.
Jan