Hello Infinia
Can that network be described as some classified compensation scheme, like NMCNR or something? Thanks.
You moved the zero gain crossover point up significantly, and you
need to consider that the caps on the drivers significantly load the
VAS at HF. I expect that my mods will not alter the slew rate, or
component sensitivity, but do significantly improve stability.
Not sure what you mean by NMCNR Salas
Rather than just adding some more poles and zeros. I thought it makes sense to find the cause of some rapid phase shifts giving some concern regarding PM and GM.
I started looking at stages individually to find the source of parasitic poles at 2-3 MHz. I looked at the output buffer stage transfer function and as sort of a test, prematurely increased the Cm value of caps at the driver.
This was an intermediate and not a suggested compensation fix. Another test was to change the driver transistor to a perhaps lower ft and power and seemed to fix the under damped resonance problem. that's all.
Not sure what to say to this Pete?
You want to be a one man band now? If so, I will be a poster from the peanut gallery throwing stones at your work. So be it.
I don't believe YOU have a valid spice model for a GFA-555!
The driver MJE150XX is not even close to the stock Toshiba parts either, let alone the outputs. More oranges to apples.
There is a scheme of Nested Miller Compensation with a an added Nulling Resistor to eliminate a zero or to move it higher. But not to perform pole zero cancellation. If it gives a +/- 50% headroom of varying the resistor's value without inflicting stability seriously then it works like NMCNR.
But now I look to your .asc and I see no nulling resistor and no take off from a step circuit's middle point to the second cap, so its just you manipulated the Nested Miller Compensation values to experimentally maybe tune better for stages and models. OK, got it. Thanks for expanding on the reasoning.
No doubt that the amp was designed in respect
of the devices specs...
I did had results quite satisfying using the original schematic,
provided the models are close enough of the one used in the amp.
The power devices are somewhat slow, 5 mhz Ft , so i took
the MJL21193/21194 which are in the same range.
The drivers have a 100mhz Ft, so it s quite logical to
keep the 2SA1837/2SC4793 which are not too much innacurate.
For the Vas, the good old 2N5401 has the same Cob as the original
device, along with a Ft tht is close.
So far, these kind of trannies seems the best to have a GFA
spice model that seems to correlate with his designers guidelines.
That said, i find that in time domain, the Onsemi power device
mentioned above are much inferior to fast devices, but then, with
fast ones, there s the problems of stability and unsatisfying
gain/phase response at high frequency...
Yes, this is the problem, wannabe experts who just throw stones to provoke the OP so that they can learn more. So far, I have provided the original file, the correct loop gain file, and the open loop file. My stability mods are reasonable, others offered here are not. And before any more cheap shots come toward my work, remember there is no charge for this work and I am not putting a lot of time into it.
So, you could ask nicely to learn more, but online people like to carelessly throw stones so that they can appear to be the informed critic. Based on what I've seen of your work so far, I don't think you've done much SPICE work. I've done this professionally for a LONG time.
It is this style of throwing stones rather than being polite that makes me hesitate every time I offer something here.
Cheap shot, I already said that I am looking for better output models.
I have already demonstrated that there is little sensitivity to the input devices.
The MJE15032/33 are the factory authorized replacement part and they
will be going into my amp when I work on it because I don't like parts
with poor SOA in the amp. So, I'm set on front end and driver
transistors, and I'll be coming up with a solution for the outputs at
some point. One of the board's SPICE experts said that they checked
out the MJE15032/33 and that they looked reasonably good.
Data for Outputs
From: Datasheet.org.uk - 2sb554 datasheet - 2sb554 application note - 2sb554 pdf
2SB554 data sheet file was too large to upload here.
From: Datasheet.org.uk - 2sb554 datasheet - 2sb554 application note - 2sb554 pdf
2SB554 data sheet file was too large to upload here.
Hi mlloyd1,
Thanks for your interest!
I started this sim with the intent of looking for more weaknesses in the Adcom design and got sidetracked with the AC analysis here, then lost interest.
Do you know of a source for good models?
I'm getting interested in looking for a simple way to do a sanity check on many of these SPICE models that seem to be so far off. I wouldn't mind a model that is at least "in the ballpark" for DC analysis rather than those that give wildly incorrect answers as I saw with the JE-990 output biasing.
Andy's models are quite good, but are not at very high IB when trying to simulate an amp with a short circuited output.
Do you have any good sources for models?
Thanks for your interest!
I started this sim with the intent of looking for more weaknesses in the Adcom design and got sidetracked with the AC analysis here, then lost interest.
Do you know of a source for good models?
I'm getting interested in looking for a simple way to do a sanity check on many of these SPICE models that seem to be so far off. I wouldn't mind a model that is at least "in the ballpark" for DC analysis rather than those that give wildly incorrect answers as I saw with the JE-990 output biasing.
Andy's models are quite good, but are not at very high IB when trying to simulate an amp with a short circuited output.
Do you have any good sources for models?
Hi PB (Panayotis).
Glad to hear you after so long time
. How are you doing?
Well, for 1 month i tried the Multisim. It is a great tool with very good models for a lot of active and passive parts. You know this probably. First time in my life, i implemented a circuit extracted from a simulator directly to a real model. It is a discrete operational amplifier which of the picture is in my web-page. To my big surprise, all measurements obtained from Multisim was in accordance (by 99%!) with those obtained from the actual model with real instruments.
Greetings from your fatherland.
Fotios
Glad to hear you after so long time
. How are you doing?Well, for 1 month i tried the Multisim. It is a great tool with very good models for a lot of active and passive parts. You know this probably. First time in my life, i implemented a circuit extracted from a simulator directly to a real model. It is a discrete operational amplifier which of the picture is in my web-page
. To my big surprise, all measurements obtained from Multisim was in accordance (by 99%!) with those obtained from the actual model with real instruments.Greetings from your fatherland.
Fotios
Hi PB (Panayotis).
Glad to hear you after so long time
Well, for 1 month i tried the Multisim. It is a great tool with very good models for a lot of active and passive parts. You know this probably. First time in my life, i implemented a circuit extracted from a simulator directly to a real model. It is a discrete operational amplifier which of the picture is in my web-page
Greetings from your fatherland.
Fotios
I'm fine Fotios, thanks and you?
Yes, I often hear good things about Multisim.
Even LTSpice works very well once the models are sorted out.
I think for example that my AC stability analysis here is probably
in the ball park.
So did you just do the 1 month eval of Multisim or do you still have it?
You had an issue with a model for the MJE182 did you ever find a
good one?
All the best,
Last edited:
@PB2
Thanks Pete for the reply. I am so fine, as this is possible in Greece of today. You hear the news for our economic crisis, and you may know that i am self-employed and NOT a state employee (the last they have resolved their economic problem enough years ago)
No, i had the eval version of Multisim. But during this time, i had as well the opportunity to transfer and store in my HDD enough Spice models. Between them, there are of course and those of MJE172 - 182 which then i imported in the Spice simulator of my own software EDWinXP v1.50. The Multisim models are very good, at least far better from those of OnSemi.
I wish you a nice weekend.
Fotios
Thanks Pete for the reply. I am so fine, as this is possible in Greece of today. You hear the news for our economic crisis, and you may know that i am self-employed and NOT a state employee (the last they have resolved their economic problem enough years ago
) No, i had the eval version of Multisim
. But during this time, i had as well the opportunity to transfer and store in my HDD enough Spice models. Between them, there are of course and those of MJE172 - 182 which then i imported in the Spice simulator of my own software EDWinXP v1.50. The Multisim models are very good, at least far better from those of OnSemi.I wish you a nice weekend.
Fotios