I highly doubt that there s as much as 58.9 db of NFB
once the frequency is higher than a few hundreds HZ...
Right, Andrew..
With the components used in Greg s try, open loop gain is at about 75.5 db.
Replacing the BD139 Vas by a 2SC3423 increase it slightly to 80.5db.
Open loop -3db frequency point is at 1.2 khz...
All in all, good values considering the simplicity of the circuit.
No wonder that this kind of schematic is overhelmingly
the most used.
With the components used in Greg s try, open loop gain is at about 75.5 db.
Replacing the BD139 Vas by a 2SC3423 increase it slightly to 80.5db.
Open loop -3db frequency point is at 1.2 khz...
All in all, good values considering the simplicity of the circuit.
No wonder that this kind of schematic is overhelmingly
the most used.
Hi Andrew,
You're on the money. If the closed loop gain is reduced, then loop gain will increase, and more lag compensation will be required to drop the gain to unity by the HF pole. Thus, for comparable stability, lag comp should be increased slightly.
Alternatively, if the closed loop gain is increased, loop gain decreases, and less lag comp is necessary to bring the loop to heel by the HF pole.
The example given was just that; a ficticious amp, not the circuit to which Wahab refers. Furthermore, I have not even mentioned the drop of loop gain with frequency, although of course it is clear that it must drop if we collar the VAS in order to bring loop gain to unity by the HF pole.
Not sure about the 'sound' of the amp. Sounds a bit subjective for you, Andrew? The 'sound' of the amp to me is more related to topology, operating point and capacitor choices.
This stability thang goes to the nub of amp design. Real fancy stuff.....
Hugh
You're on the money. If the closed loop gain is reduced, then loop gain will increase, and more lag compensation will be required to drop the gain to unity by the HF pole. Thus, for comparable stability, lag comp should be increased slightly.
Alternatively, if the closed loop gain is increased, loop gain decreases, and less lag comp is necessary to bring the loop to heel by the HF pole.
The example given was just that; a ficticious amp, not the circuit to which Wahab refers. Furthermore, I have not even mentioned the drop of loop gain with frequency, although of course it is clear that it must drop if we collar the VAS in order to bring loop gain to unity by the HF pole.
Not sure about the 'sound' of the amp. Sounds a bit subjective for you, Andrew? The 'sound' of the amp to me is more related to topology, operating point and capacitor choices.
This stability thang goes to the nub of amp design. Real fancy stuff.....
Hugh
Last edited:
I'm not too good with this stabilty design and don't seem to be getting any better for all the advice that members have given to me.
But when Cdom (D. Self speak) is changed you can easily hear the effect and it is measurable. The oscilloscope shows the change admirably.
If the NFB ratio is changed, that too shows up, but when the corresponding Cdom correction is made, I find that the amp often sounds different. I never seem able to get back to the designers sound.
To me it seems that it's not just the dominant pole that needs to be set correctly, it's something to do with open loop gain and NFB ratio and probably lot's of other things, but the big effect seems to be the NFB feedback ratio.
But when Cdom (D. Self speak) is changed you can easily hear the effect and it is measurable. The oscilloscope shows the change admirably.
If the NFB ratio is changed, that too shows up, but when the corresponding Cdom correction is made, I find that the amp often sounds different. I never seem able to get back to the designers sound.
To me it seems that it's not just the dominant pole that needs to be set correctly, it's something to do with open loop gain and NFB ratio and probably lot's of other things, but the big effect seems to be the NFB feedback ratio.
Last edited:
Differences in amplifier sound when changing Cdom may well be down to the fact that when doing this you are changing the slew rate and the loop gain unity gain frequency - assuming that the LTP tail current is held constant. I have not tried a direct comparison, but I would suspect that in a slew rate limited deisgn like the baby AKSA, the sonic differences will be quite easy to pick out. In a non-slewing design, this is probably going to be a bit more difficult.
Good point ,Bonsai.
The slew rate of this amp is about 4.5V/uS , so at high level,
it will even enter in slew rate limitation.
As pointed by Hugh elsewhere in this thread , the compensation
capacitor must be drastically reduced.
The 22pF value he s quoting is the minimal value while
retaining a sufficent stability.
27 pF is still largely enough to improve stability a little
while retaining a sufficent SR of 11V/uS.
Hi Bonsai,
Thanks for the input.
These are the difference between the B-AKSA and "real" AKSA. The tail current on the B-AKSA is different and the other tweaks are missing.
Hugh would not be happy if I published the exact circuit with exact values.
.regards
Last edited:
Hi Bonsai,
Thanks for the input.
These are the difference between the B-AKSA and "real" AKSA. The tail current on the B-AKSA is different and the other tweaks are missing.
Hugh would not be happy if I published the exact circuit with exact values.
regards
OK then, thats an invitation to 'tweak it' and see what you come up with!
Member
Joined 2009
Paid Member
Hugh I am glad you mentioned the Self EF II output topology.
After looking at D. Self's excellent book I noticed that the suckout cap across the bases of the output transistors (Q7, Q8) is a component that is missing from the B-AKSA schematic. I was looking for an excuse to introduce it and by an odd co-incidence there are 2 test points (TP3, TP4) on the PCB that could be used.
I have seen the similar circuits with the suckout cap in 3 locations:
1. Parallel to the resistor (R15) across the emitter of the drivers (Q5, Q6)
2. Connected to the O/P base resistors (R16, R17) on the driver (Q5, Q6) side (electrically the same as above)
3. Directly to the base of the output transistors (Q7, Q8)
Is there a preference? I though it was an aid to the output transistors and as such, having the cap work through the base resistors would only slow it down.
If there was no base resistors (R16, R17) then all 3 would options would be identical.
I have attached the schematic again for reference.
regards
Hi Greg, just 2c worth
The work of reversing the OP base charges is done by the drivers so wherever you place the capacitor, there will be some limit due to R16 & 17 on the charge removal.
The average lowest impedance, straight across the bases may be best.
Dare I suggest.......simulation?
Randy Slone(if I could find his book here) discussed this too, hinting at an R/C time factor in the charge removal, as I recall.
The work of reversing the OP base charges is done by the drivers so wherever you place the capacitor, there will be some limit due to R16 & 17 on the charge removal.
The average lowest impedance, straight across the bases may be best.
Dare I suggest.......simulation?
Randy Slone(if I could find his book here) discussed this too, hinting at an R/C time factor in the charge removal, as I recall.
I've got it in LTspice, I'm not sure about the accuracy of the models for the transistors, but if anyone wants it I can upload the asy file
Tony.
Hi Ian, you mean 5c don't you.
After reading through Self and Slone a few times, I have to admit the process is still not 100% clear, but I am starting to wonder if the base stoppers are worth having.
I'll will have a go at simulating this.
EDIT: I have this in LTspice Tony but can't guarantee the models either, plus I am bad at driving LTspice.
regards
Last edited:
Me too Greg, but getting better I found it a fantastic tool for trying stuff out, and trying to get a better idea of how something works. Fantastic for just trying random changes to see what happens, without the fear of lifting a track or heat damaging a component. Though I haven't made the leap from simulating to building yet to find out if it actually works!! Slowly Slowly
I obsess over things for way to long. though I think I've almost settled on the components I want for my passive crossover, so then I can put in the order for the other stuff I need for the active one, and also some parts for the baby aksa as well... I will get there
Tony.
I found it a fantastic tool for trying stuff out, and trying to get a better idea of how something works. Fantastic for just trying random changes to see what happens, without the fear of lifting a track or heat damaging a component. Though I haven't made the leap from simulating to building yet to find out if it actually works!! Slowly Slowly I obsess over things for way to long. though I think I've almost settled on the components I want for my passive crossover, so then I can put in the order for the other stuff I need for the active one, and also some parts for the baby aksa as well... I will get there
Tony.
Tell me about it. Now which smilie icon should I use, the green one or the yellow one, or there might be an even better one if I select [More].....
I have that particular doubt when I look at the schematic and those resistors. Then I don't have engineers disdain for meaningless blots on paper and they stick in my tiny brain. I have to see them as more than just 10 ohms and just a small percentage of R15 or whatever.
This is a big problem and yes, you cant give 2c worth anymore...darn 5c rounding!
I never thought to ask where those stoppers came from...sim? perhaps I shouldn't ask either...(secrets...voodoo etc.)
This is a big problem and yes, you cant give 2c worth anymore...darn 5c rounding!
I never thought to ask where those stoppers came from...sim? perhaps I shouldn't ask either...(secrets...voodoo etc.)
Last edited: