Hi guys I haven't post anything for a long time but I still working on this project I just want to say thank you guys for all the tips helps on my building specially mister Destroyer X, I have plans to order a set of board for myself in case a have a mayor burn out lol, well guys have a wonderful day.
Regards
vargasmongo3435
Regards
vargasmongo3435
Changing the Feedback Ratio
I would also like to reduce the amps gain to something closer to 20X. My reading suggests that increasing the global feedback will also require a change in Cdom. Although I've located the formula for calculating the correct value for Cdom it requires knowlege of the amps closed loop gain bandwidth and the transconductance of the input LTP. I didn't find these parameters available anywhere within the forum but perhaps someone knows what they are or how to derive them from the schematic. I'm concerned that simply increasing the negative feedback without adjusting Cdom might lead to instability.
I would also like to reduce the amps gain to something closer to 20X. My reading suggests that increasing the global feedback will also require a change in Cdom. Although I've located the formula for calculating the correct value for Cdom it requires knowlege of the amps closed loop gain bandwidth and the transconductance of the input LTP. I didn't find these parameters available anywhere within the forum but perhaps someone knows what they are or how to derive them from the schematic. I'm concerned that simply increasing the negative feedback without adjusting Cdom might lead to instability.
Interresting point bonfis. You seem right saying that by changing NFB we should reconsider the miller cap value. Because I barely start understanding its role in the schematic, I decided to compare DX amps to identify similarities. Here are my findings
Here are the gain vs Cdom values of the amps mr. DX has developped:
ES = 26.5x (28dB) = 82pF
ST = 26.5x (28dB) = 100pF
MKII = 31x (30dB) = 100pF
MKII-SC = 58x (35dB) = 100pF
HX = 121x (41dB) = 180pF
The "default" value Dr. Self uses for his amps is of 100pF.
I decided to simulate my schematic and compare THD at near-full power at 20Khz.
Simulation details:
Gain= 32x (30dB)
Frequency = 20kHz
Output power = 200W RMS / 8ohm resistive load
Input = 1.244V RMS.
with 180pF Cdom = 0.063% THD
With 100pF Cdom = 0.039% THD
With 82pF Cdom = 0.033% THD
Interresting... but is it stable with lower values?... I don't have a scope to test the real thing
Only my ears, and they tell me that the amp sounds good with 180pF, but can it be good to lower that value to 100pF?.
share your findings!
Mart.
Here are the gain vs Cdom values of the amps mr. DX has developped:
ES = 26.5x (28dB) = 82pF
ST = 26.5x (28dB) = 100pF
MKII = 31x (30dB) = 100pF
MKII-SC = 58x (35dB) = 100pF
HX = 121x (41dB) = 180pF
The "default" value Dr. Self uses for his amps is of 100pF.
I decided to simulate my schematic and compare THD at near-full power at 20Khz.
Simulation details:
Gain= 32x (30dB)
Frequency = 20kHz
Output power = 200W RMS / 8ohm resistive load
Input = 1.244V RMS.
with 180pF Cdom = 0.063% THD
With 100pF Cdom = 0.039% THD
With 82pF Cdom = 0.033% THD
Interresting... but is it stable with lower values?... I don't have a scope to test the real thing
Only my ears, and they tell me that the amp sounds good with 180pF, but can it be good to lower that value to 100pF?.share your findings!
Mart.
About the miller capacitor, Destroyer X said in his video Monologue detailing the MKIII-Hx power audio amp:
"...I have used in some schematics 82pF, yes you can use that value if you want but it is dangerous, better for you to keep it 180pF because it is tricky to check if it's oscillating you need a scope to see ocillations trigerred by the signal..."
Then later on in the same video:
"...do not change it, keep this damn thing 180pF, it's gonna be safer this way..."
So to my understanding, lowering the value can elevate the risk of oscillations.
But what is the relationship between the gain and the Cdom value?
"...I have used in some schematics 82pF, yes you can use that value if you want but it is dangerous, better for you to keep it 180pF because it is tricky to check if it's oscillating you need a scope to see ocillations trigerred by the signal..."
Then later on in the same video:
"...do not change it, keep this damn thing 180pF, it's gonna be safer this way..."
So to my understanding, lowering the value can elevate the risk of oscillations.
But what is the relationship between the gain and the Cdom value?
Hi Mart-
Thanks very much for the simulation results. It does seem that for the Blame MK III the distortion results benefit from a reduction in Cdom when feedback is increased. The formula for calculating Cdom (from Bob Cordell's book) indicates that the correct value for Cdom is inversely proportional to closed loop gain if the input stage transconductance and closed loop bandwidth remain constant. So a reduction in the closed loop gain should require an increase in the value of Cdom (everything else being equal). Perhaps the only consequence to be considered when calculating Cdom is stability and not distortion. However, since global feedback is being increased in the process of reducing overall gain I would expect distortion results might improve also.
Steve
Thanks very much for the simulation results. It does seem that for the Blame MK III the distortion results benefit from a reduction in Cdom when feedback is increased. The formula for calculating Cdom (from Bob Cordell's book) indicates that the correct value for Cdom is inversely proportional to closed loop gain if the input stage transconductance and closed loop bandwidth remain constant. So a reduction in the closed loop gain should require an increase in the value of Cdom (everything else being equal). Perhaps the only consequence to be considered when calculating Cdom is stability and not distortion. However, since global feedback is being increased in the process of reducing overall gain I would expect distortion results might improve also.
Steve
Cdom or Miller cap, is there to maintain stability.
Set it so that the amplifier is stable.
Increasing the value of Cdom reduces the open loop gain of the amplifier by increasing the local feedback around around the VAS. That reduction of Open Loop Gain reduces the feedback in the global feedback loop and that is what increases the stability margins.
Cordell is correct. Self is correct.
As a consequence of setting the Cdom value as low as possible and still maintaining the required stability, the open loop gain is set to maximum. With the closed loop gain fixed, that results in more feedback. That increased feedback reduces all the various characteristics that rely on feedback. Distortion is just one of them.
In summary.
reducing Cdom reduces the stability margins.
reducing Cdom reduces distortion.
Set it so that the amplifier is stable.
Increasing the value of Cdom reduces the open loop gain of the amplifier by increasing the local feedback around around the VAS. That reduction of Open Loop Gain reduces the feedback in the global feedback loop and that is what increases the stability margins.
Cordell is correct. Self is correct.
As a consequence of setting the Cdom value as low as possible and still maintaining the required stability, the open loop gain is set to maximum. With the closed loop gain fixed, that results in more feedback. That increased feedback reduces all the various characteristics that rely on feedback. Distortion is just one of them.
In summary.
reducing Cdom reduces the stability margins.
reducing Cdom reduces distortion.
Last edited:
Great explanations, this makes things clearer around that oh-so important cap.
So for people like me that likes to try things and doesn't necessarily have all the required instrumentations, what could be the test path?
I think I can refer to other DX designs, their gain and Cdom value to try and determine a new value for Cdom. I know that DX has tested with 121x gain and 56pF Cdom, but it WAS unstable.
I guess that from 121x gain and 180pF, going down to 32x gain permits going to 100pF without any problem.
But without instrumentation, what is the indicator of oscillations? Will wi hear anything wrong? will the amp self-destroy?
From my knowledge, a way to monitor is to check the temperature of the 15R as shown below, which is the escape route for the oscillations... (see pic)... Any thoughts?
Regards,
Mart.
So for people like me that likes to try things and doesn't necessarily have all the required instrumentations, what could be the test path?
I think I can refer to other DX designs, their gain and Cdom value to try and determine a new value for Cdom. I know that DX has tested with 121x gain and 56pF Cdom, but it WAS unstable.
I guess that from 121x gain and 180pF, going down to 32x gain permits going to 100pF without any problem.
But without instrumentation, what is the indicator of oscillations? Will wi hear anything wrong? will the amp self-destroy?
From my knowledge, a way to monitor is to check the temperature of the 15R as shown below, which is the escape route for the oscillations... (see pic)... Any thoughts?
Regards,
Mart.
Attachments
Wrong guess.
As one reduces the closed loop gain, one is also increasing the feedback, making the amplifier less stable. To correct for this one would expect the Cdom to get higher in value as the closed loop gain is reduced.
What Andrew says agrees with the formula in Cordell's book. Cdom is inversely related to closed loop gain. So if the amp is stable even with the increased feedback and original 180pf then there would be no need to increase Cdom. The question would be do you want to risk decreasing it.
For a scope free stability test Cordell also suggests looking at the closed loop frequency response for evidence of any peaking (anymore than 1 db) just before the final roll-off. Any peaking is a "danger sign." He notes however that absence of peaks doesn't guarantee stability. You need to bypass any input low pass filters while running this test.
He also describes tests that can be run in simulation but they are too complicated to include here.
I see. I've taken temp readings on the VAS, the CCS and the output trannys when I reduced the global closed-loop gain, and everything stayed the same. I hope this indicates it's stable.
With my little knowledge, it would be too risky to blindly tamper with Cdom. I will wait until I understand better its role and method of calculation before changing it.
In the meantime maybe someone will make some real-world measurements and post his results. We miss DX's knowledge right here...
With my little knowledge, it would be too risky to blindly tamper with Cdom. I will wait until I understand better its role and method of calculation before changing it.
In the meantime maybe someone will make some real-world measurements and post his results. We miss DX's knowledge right here...
- Status
- Not open for further replies.