Thanks to Twest for throwing in the term of input related.
My previous was pretty much misleading.
Input related noise:
Here it does not help to increase the loop gain for reducing the noise, but it is hrlpful to reduce the closed loop gain and use larger input signals.
Errors which are not input related:
They will not be reduced by changing the feedback in a way which reduce the closed loop gain while keeping the loop gain unchanged.
Here only more loop gain in the concerned frequency range will help.
Why does loop gain not automatically grow by reducing the closed loop gain?
Your proposed way of reducing the closed loop gain, will also reduce the open loop gain (however not by the exactly identical amount).
Stable also with 22k input resistor?
I guess yes, but with less margin.
AFEC:
I have no opinion. The article just tells that it is old and great. Furtheron former engineers are blamed to have been unable to calculate such complicated things.
Not really helpfull in terms of control theory.
My previous was pretty much misleading.
Input related noise:
Here it does not help to increase the loop gain for reducing the noise, but it is hrlpful to reduce the closed loop gain and use larger input signals.
Errors which are not input related:
They will not be reduced by changing the feedback in a way which reduce the closed loop gain while keeping the loop gain unchanged.
Here only more loop gain in the concerned frequency range will help.
Why does loop gain not automatically grow by reducing the closed loop gain?
Your proposed way of reducing the closed loop gain, will also reduce the open loop gain (however not by the exactly identical amount).
Stable also with 22k input resistor?
I guess yes, but with less margin.
AFEC:
I have no opinion. The article just tells that it is old and great. Furtheron former engineers are blamed to have been unable to calculate such complicated things.
Not really helpfull in terms of control theory.
Last edited:
Hi,
So, instead we reduce the Feedback resistor to 5.6k. Can we do that without throwing the secondary loop that establishes oscillation totally out? Will it be stable? Do we get less THD relating to internal nonlinearity and less noise that way.
Sorry to ask such dumb questions.
As for AFEC, I may be wrong but ain't that what the Hypex NCore uses?
Sir S
So, instead we reduce the Feedback resistor to 5.6k. Can we do that without throwing the secondary loop that establishes oscillation totally out? Will it be stable? Do we get less THD relating to internal nonlinearity and less noise that way.
Sorry to ask such dumb questions.
As for AFEC, I may be wrong but ain't that what the Hypex NCore uses?
Sir S
No, this for sure will completely mess the frequency of oscillation and the entire loop gain adjustment.
My recommendation:
Read this thread carefully. Download the provided simulations.
Experiment with the simulations. The fit between the results of the attached simulations and reality is reasonably good. Only when the simulation is running fine, then it is making sense to try the mod in reality.
However regarding THD generally such simple simulations are not suited to judge distortion levels below -70db. ...figures like the real LiteAmp does provide, definitely need detail optimization at the real prototype.
Hello!
After so many weeks I found again some time to soldering and finish my LiteAmp. Yesterday it could have been that day. But....
Main Board: At Get it going I was probably already tied and at points 1 and 2 I've done some wrong wiring and manage to put my smps on silent at point 4. After that I discovered the error and put the right wires but at point 4 (Power up) my auxiliary resistor from point 3 start immediately to smoke.
What could be wrong now?
My mind is empty...
Thank you!
After so many weeks I found again some time to soldering and finish my LiteAmp. Yesterday it could have been that day. But....
Main Board: At Get it going I was probably already tied and at points 1 and 2 I've done some wrong wiring and manage to put my smps on silent at point 4. After that I discovered the error and put the right wires but at point 4 (Power up) my auxiliary resistor from point 3 start immediately to smoke.
What could be wrong now?
My mind is empty...
Thank you!
Do you remember exactly which error connection you had made?
This would help to guess which defects it might have caused.
In case the connections of point 1&2 are in place, then there can be two reasons why this additional resistor might smoke.
a) Do you have connected a load resistor? If yes, then the small additional resistor wil get overloaded. All tests before point 23 are intended to be done without load resistor. Sorry that I did not highlight this in the instructions.
b) Potentially your wiring error might have caused a defect MosFet (or more...). Check and if necessary exchange the MosFet(s).
If you are short on MosFets: The amp basically also will work with just one of the IRFI, however dead time adjustment will not be perfect anymore (potentially some more heat during idle) and cannot support 2R operation any more. But a fundamental get it going is possible with just one IRFI.
You will need the diode testing function of your multimeter.
In each IRFI there are two MosFets, connected to a halbridge.
Means you will find two diode paths.
Here a describtion how to check MosFets.
There are many more descriptions in the WEB, but this was one of the first hits of the search engine.
Testing a MosFet
In each IRFI there are two MosFets, connected to a halbridge.
Means you will find two diode paths.
Here a describtion how to check MosFets.
There are many more descriptions in the WEB, but this was one of the first hits of the search engine.
Testing a MosFet
D
Deleted member 148505
One Mosfet was gone (i think it was the Q6) and the other one seems to be OK measured like this. I've put a new one in the Q6 place with Q7 empty and all seems to be OK except pin 6 to GND that is -3.35VDC. I've check the R2, R3, R4, R5, D2, C9, Q2 and it wal all OK. What to do next?
Thank you for your patience.
Thank you for your patience.
Is your IRS2092 already inserted? A defect 2092 could cause this.
If IRS2092 is already inserted: Remove and check and if Ok - go on with a new 2092.
If IRS2092 is not inserted: Check Q3 and Q4. Also defect Q3 and/or Q4 (Q4 more likely) could cause -3.35V.
I've replaced Q3 with a new one. I have to buy Q2, Q5, Q4 on Monday. I checked again and I think both Q2 and Q5 are defective.
IRS2092 was not inserted. I followed the steps but simply I messed something in step 1 and 2. On both boards. Well one is healthy now but for the moment I can't go to step 16 - Insert U1. Because... I realize that I've ordered IRS2092STRPBF instead of IRS2092PBF. And I learned something about SOIC-16 and DIP-16
. Mine is a lot smaller and I can't use it I think. I will order the DIP-16 IRS2092PBF. Looking twice on the pictures with the finished board I see that you insert it in something black with 16 holes. Does it comes with the chip or I have to order it separately?
And one more. In that night I was about to finish the gain board either. Maybe that's why later the confusion. I also realize than that I didn't order C110 because I couldn't find one with Leakage < 0.5uA. Do you have any part number for example?
I am sure that Monday I will fix the board. Thank you all for your help.
IRS2092 was not inserted. I followed the steps but simply I messed something in step 1 and 2. On both boards. Well one is healthy now but for the moment I can't go to step 16 - Insert U1. Because... I realize that I've ordered IRS2092STRPBF instead of IRS2092PBF. And I learned something about SOIC-16 and DIP-16
. Mine is a lot smaller and I can't use it I think. I will order the DIP-16 IRS2092PBF. Looking twice on the pictures with the finished board I see that you insert it in something black with 16 holes. Does it comes with the chip or I have to order it separately?And one more. In that night I was about to finish the gain board either. Maybe that's why later the confusion
. I also realize than that I didn't order C110 because I couldn't find one with Leakage < 0.5uA. Do you have any part number for example?I am sure that Monday I will fix the board
. Thank you all for your help.This something black with 16 holes is a DIL/DIP16 socket of cheapest most ordinary style. It is not coming with the IRS2092, but it is not necessary. It just makes it easier
to replace the IRS2092. During R&D I am most often using sockets, except HF-requirements would not allow - but here it is easily allowed.
Have a look to posting #395
http://www.diyaudio.com/forums/class-d/255046-systemd-liteamp-40.html#post4430164