Hi,
just realised I didn't mention the burst repeat timescale.
About 25uS to 30uS from burst start to start. Each burst lasts about 15uS to 20uS.
just realised I didn't mention the burst repeat timescale.
About 25uS to 30uS from burst start to start. Each burst lasts about 15uS to 20uS.
Hi Andrew,seems like my scope picked up something afterall.It's only a dual 15Mhz memory scope but with the delayed screen on i could clearly see a modulated 5MHz wave.The modulation looked almost like a sine wave on my scope.The 5MHz was frequency stable but the voltage was modulated from 0 to 2mV in a sine wave pattern.I took the full cycle time and came up with 130kHz.Perhaps this changes with input voltage or voltage drop across the regulator?
Kees.
Kees.
Andrew and Kees,
If you have a chance could you try increasing C7 or decreasing R8 to see if that elminates the 5 MHz bursts?
I cannot see it if it is there on my supplies. The bursts that I had were much lower frequency and at intervals related to diode switching.
If you have a chance could you try increasing C7 or decreasing R8 to see if that elminates the 5 MHz bursts?
I cannot see it if it is there on my supplies. The bursts that I had were much lower frequency and at intervals related to diode switching.
Hi,
I checked the negative reg to see if same oscillation. Bursts go in opposite direction otherwise similar.
Today, I saw a distinct slope in the space between bursts. But sometimes only continuous 5MHz @ 6mV to 7mVpp when Vin=50Vdc. Don't know what causes the bursts to apparently change to continuous.
Tried R8=1k5, no significant change.
Tried R8=1k5 and C7=22pF, bursts less easy to lock on to.
Difficult to measure/see what's happening. Peak to peak about the same.
But, it is quite clear that when Vin is low and I lose regulation the oscillation disappears, at about Vt1=1V then 2mVpp, at Vt1=8V then osc=7mVpp.
After the motoGP I'll try a larger C7.
I checked the negative reg to see if same oscillation. Bursts go in opposite direction otherwise similar.
Today, I saw a distinct slope in the space between bursts. But sometimes only continuous 5MHz @ 6mV to 7mVpp when Vin=50Vdc. Don't know what causes the bursts to apparently change to continuous.
Tried R8=1k5, no significant change.
Tried R8=1k5 and C7=22pF, bursts less easy to lock on to.
Difficult to measure/see what's happening. Peak to peak about the same.
But, it is quite clear that when Vin is low and I lose regulation the oscillation disappears, at about Vt1=1V then 2mVpp, at Vt1=8V then osc=7mVpp.
After the motoGP I'll try a larger C7.
Hi,
R8=1k5, C7=1nF, osc=5MHz 22mVpp continuous @Vt1=8V, Vin 51.6Vdc.
C7 from 22pF, thro' 100pF to 1000pF, definitely worse at 1000pF, maybe better at 22p but difficult to lock in.
If you want to watch the GP later, then look away now.
Rossi has closed the gap to 29points.
Since the circuit is working it does not need current protection so next step is to try transformer feed in case the lab supply is aggravating/causing the problem.
R8=1k5, C7=1nF, osc=5MHz 22mVpp continuous @Vt1=8V, Vin 51.6Vdc.
C7 from 22pF, thro' 100pF to 1000pF, definitely worse at 1000pF, maybe better at 22p but difficult to lock in.
If you want to watch the GP later, then look away now.
Rossi has closed the gap to 29points.
Since the circuit is working it does not need current protection so next step is to try transformer feed in case the lab supply is aggravating/causing the problem.
Thanks for the efforts Andrew.
Interesting that the Oscillation gets worse with larger C7. I would have figured that larger C7 would limit the bandwidth and kill the oscillation.
If smaller C7 seems better, how about getting rid of it all together? What do you think Jens?
Interesting that the Oscillation gets worse with larger C7. I would have figured that larger C7 would limit the bandwidth and kill the oscillation.
If smaller C7 seems better, how about getting rid of it all together? What do you think Jens?
Hi,
I connected 38Vac and tested my DIY oscillator. I give up.
I went back to the original low voltage board and connected 16.5Vac, Vout 18Vdc. It is a lot better.
Noise of about 2mVpp
5MHz oscillation, no bursts, just continuous, about 1mVpp.
when the regulation fails due to too high an output voltage I see the flat topped ripple expected, but the dips as the output tracks the ripple are completely filled out with oscillation.
I think the bursts I saw earlier were an interaction with the control circuits in the lab supply.
All you designers out there, come up with a solution.
I connected 38Vac and tested my DIY oscillator. I give up.
I went back to the original low voltage board and connected 16.5Vac, Vout 18Vdc. It is a lot better.
Noise of about 2mVpp
5MHz oscillation, no bursts, just continuous, about 1mVpp.
when the regulation fails due to too high an output voltage I see the flat topped ripple expected, but the dips as the output tracks the ripple are completely filled out with oscillation.
I think the bursts I saw earlier were an interaction with the control circuits in the lab supply.
All you designers out there, come up with a solution.
Hi Bob,
making C7 bigger makes the amp slower, I would expect that to make the whole (amp & output) more likely to oscillate. That's why I tried a smaller C7 first to bring up the F-3 point to near 5MHz. and use the lower R8 to damp sudden changes. Just shows how little I know.
making C7 bigger makes the amp slower, I would expect that to make the whole (amp & output) more likely to oscillate. That's why I tried a smaller C7 first to bring up the F-3 point to near 5MHz. and use the lower R8 to damp sudden changes. Just shows how little I know.
I tried increasing C7 to 1nF with no results.Tried bypassing R10 with 100nF ,The 130kHz got much smaller but the 5MHz stayed.
Anyway, with my scope it is difficult to lock on to the signal which means constantly adjusing the manual trigger and keeping the probes in the right place to measure different points.
As you said Bob, a small cap to ground should kill such a small 5MHz signal.Why it doesn't i don't know.
Kees.
Anyway, with my scope it is difficult to lock on to the signal which means constantly adjusing the manual trigger and keeping the probes in the right place to measure different points.
As you said Bob, a small cap to ground should kill such a small 5MHz signal.Why it doesn't i don't know.
Kees.
Well this has me buffaloed (completely confused). Since I don't have/can't see the 5 MHz stuff, I can't experiment and try to solve it.
Andy's now acting more like mine was with long leads and low AC input voltage. For me the oscillation was on the flat tops of the ripple, though. Maybe an RC snubber on the input would help. I don't know where to start with values, though.
How about trying it without C7 and/or with 10 pF or so from C to B on T2?
Hopefully Jens will be able to join this discussion.
Andy's now acting more like mine was with long leads and low AC input voltage. For me the oscillation was on the flat tops of the ripple, though. Maybe an RC snubber on the input would help. I don't know where to start with values, though.
How about trying it without C7 and/or with 10 pF or so from C to B on T2?
Hopefully Jens will be able to join this discussion.
Andrew,
I also have been unable to detect the problem here as long as I keep the inputvoltage about 5V above the output. May I suggest you try removing C7?
I have not yet build a high voltage version as I intend to use this regulator for active x-overs only.
The output cap is important for stability.
Try making C3 larger in value.
\Jens
I also have been unable to detect the problem here as long as I keep the inputvoltage about 5V above the output. May I suggest you try removing C7?
I have not yet build a high voltage version as I intend to use this regulator for active x-overs only.
The output cap is important for stability.
Try making C3 larger in value.
\Jens
Andrew,
I stand corrected, I can also see 5Mhz when the supply is loaded with 15Ohm and supplying 15V.
I missed that before as I focused on behaviour without load.
I found some 5 Mhz on the reference voltage and decided not to let R4 be 1k at high frequencies so I bypassed it with 1uF. That seemed to get rid of the oscillation.
Can you confirm these findings?
Sorry for the lack of support up to this point… I think this will do the trick.
\Jens
I stand corrected, I can also see 5Mhz when the supply is loaded with 15Ohm and supplying 15V.
I missed that before as I focused on behaviour without load.
I found some 5 Mhz on the reference voltage and decided not to let R4 be 1k at high frequencies so I bypassed it with 1uF. That seemed to get rid of the oscillation.
Can you confirm these findings?
Sorry for the lack of support up to this point… I think this will do the trick.
\Jens
Hi,
I'm going to try C7 removal & R4 cap bypass.
C7 removal:- is that open circuit or shorting link?
I'm going to try C7 removal & R4 cap bypass.
C7 removal:- is that open circuit or shorting link?
I tried bypassing R4 with a 1nF cap before with no results.I will try with a bigger cap.I will also try removing C7.
Kees.
Kees.
Andy and Kees,
Have you had a chance to do any further testing? Eagerly awaiting your results.
/Bob
Have you had a chance to do any further testing? Eagerly awaiting your results.
/Bob
Hi,
not since the weekend, but maybe before next weekend.
Another thought:- replace D6 with a cap. Overall loop HiFreq NFB.
This might allow removal of C7 (open circuit).
Jens,
could you model this?
What value of R7 would cover the range of output currents and output voltages?
Yes, I realise that D6 must be replaced (how about //R4 on back of PCB).
not since the weekend, but maybe before next weekend.
Another thought:- replace D6 with a cap. Overall loop HiFreq NFB.
This might allow removal of C7 (open circuit).
Jens,
could you model this?
What value of R7 would cover the range of output currents and output voltages?
Yes, I realise that D6 must be replaced (how about //R4 on back of PCB).
Hi all,
I did some more lab experiments and found out why I did not observe the 5 MHz oscillator like behaviour before.
When I load the regulator with 0.5A I can only get the oscillation to occur when I have low DC input voltage.
At 21V or above the oscillation does NOT occur.
If I shunt R4 with a 1µF capacitor the voltage drops to 18V
Initially I only tested using 18VAC (50VA) transformer this gives about 23.5V DC and thus the regulator behaved nicely.
I thing this might be the reason that Bob also have problems in detecting the phenomenon – Bob can you confirm this?
Anyhow the regulator IS stable as long as you feed it with enough voltage.
If you want it to run with lower input voltages please shunt R4 with a 1µF cap.
FYI: none of the problems show up in the spice model I have made for this project.
If anyone can confirm this observation – please respond.
Thanks
\Jens
I did some more lab experiments and found out why I did not observe the 5 MHz oscillator like behaviour before.
When I load the regulator with 0.5A I can only get the oscillation to occur when I have low DC input voltage.
At 21V or above the oscillation does NOT occur.
If I shunt R4 with a 1µF capacitor the voltage drops to 18V
Initially I only tested using 18VAC (50VA) transformer this gives about 23.5V DC and thus the regulator behaved nicely.
I thing this might be the reason that Bob also have problems in detecting the phenomenon – Bob can you confirm this?
Anyhow the regulator IS stable as long as you feed it with enough voltage.
If you want it to run with lower input voltages please shunt R4 with a 1µF cap.
FYI: none of the problems show up in the spice model I have made for this project.
If anyone can confirm this observation – please respond.
Thanks
\Jens
I haven't had the time to do further testing.Perhaps before the weekend.
Andrew, do you use the LM4040 as a reference?
Have you tried using a standard zenerdiode?
Kees.
Andrew, do you use the LM4040 as a reference?
Have you tried using a standard zenerdiode?
Kees.
- Status
- Not open for further replies.