The picture of my "beauty" didn't work dammit... 😛 Let's try it again:
An externally hosted image should be here but it was not working when we last tested it.
Curious to understand more about what the problem might be with the PGA23XX series. Did you listen to all the variants, or just the 'original' 2310 ? It seems from a quick look at the datasheets that they played around with the output buffer opamp between the variants. How they can claim 'high performance' for a 1MHz unity gain bandwidth (PGA2320 datasheet) is beyond me - its the same as for a 741😱
Hi,
Well, I have experience with the CS3310 and PGA2300 and 2310.
I suspect the problem is the topology, the switched resistor network in the op-amp's feedback path (exposes fet switches non-linearity) and generally the use of crappy on board op-amp's.
As said, the better Fet switched based attenuators (even with diffusion resistors) perform very well, IF (and only if) the load impedance is very high.
Ciao T
Well, I have experience with the CS3310 and PGA2300 and 2310.
I suspect the problem is the topology, the switched resistor network in the op-amp's feedback path (exposes fet switches non-linearity) and generally the use of crappy on board op-amp's.
As said, the better Fet switched based attenuators (even with diffusion resistors) perform very well, IF (and only if) the load impedance is very high.
Ciao T
Black's feedforward is not easily applied due to the required power output combiner
this ideally has to combine the 2 amp's outputs without interaction of the two output stages
frequency dependent power crossover networks can come close to doing the job but most people won't be aware of the stability problem with trying to make a "lossless" combiner - you end up needing to waste power in the combiner to avoid a cancellation zero that occurs with pure L,C crossovers
power feedforward also requires overlap in frequency and some compatibility in I,V characteristics - I'd look at using something like the 1 A output LT1210 CFA in a TO-220 instead of pdip op amps
the limited bandwidth of audio power chip amps limits the practical crossover frequency to < ~100KHz where some signal sources may still have large amplitude signals (even if arguably not "audio") - unless you measure to assure you never get stuck with the 2 MHz min GBW, +10 min stable gain LM38xx Chips
with tolerance, phase and ripple errors from the power combiner you still want to wrap the whole in a nested feedback loop, but you can apply more feedback to higher frequencies due to the faster feedforward stage – at least for linear stability – nonlinear stability in clipping/slew limiting may become very interesting with such a composite amp
this ideally has to combine the 2 amp's outputs without interaction of the two output stages
frequency dependent power crossover networks can come close to doing the job but most people won't be aware of the stability problem with trying to make a "lossless" combiner - you end up needing to waste power in the combiner to avoid a cancellation zero that occurs with pure L,C crossovers
power feedforward also requires overlap in frequency and some compatibility in I,V characteristics - I'd look at using something like the 1 A output LT1210 CFA in a TO-220 instead of pdip op amps
the limited bandwidth of audio power chip amps limits the practical crossover frequency to < ~100KHz where some signal sources may still have large amplitude signals (even if arguably not "audio") - unless you measure to assure you never get stuck with the 2 MHz min GBW, +10 min stable gain LM38xx Chips
with tolerance, phase and ripple errors from the power combiner you still want to wrap the whole in a nested feedback loop, but you can apply more feedback to higher frequencies due to the faster feedforward stage – at least for linear stability – nonlinear stability in clipping/slew limiting may become very interesting with such a composite amp
OK... Up until now I understood people's posts at least partly... But I can't understand a single sentence of jcx's post above 😛 😛
people have long wanted to combine different amp outputs, a cheap high power amp and better "quality" lower power amp to get a composite amp with both high power and "good quality"
I was addressing Thorsten's comment on the previous page about Black's error feedforward technique as an example of the 2 amp approach - Black's technique requires adding 2 amp's outputs without their loading each other - not an easy task
we've already discussed the popular paralleling of similar outputs with small value load balancing resistors, an example of a lossey combiner which also gives high loading interaction of the 2 outputs - increasing power but only indirectly improving the amp thru greater current output capability and effective decrease in load seen by each of the individual amps with shared outputs
I was pointing to the problems with some of the combiner proposals that keep popping up when you try to “improve” the output using different type amps - particularly the slow high power amp (LM38xx and relatives in this discussion) and a fast low power amp combination
I was addressing Thorsten's comment on the previous page about Black's error feedforward technique as an example of the 2 amp approach - Black's technique requires adding 2 amp's outputs without their loading each other - not an easy task
we've already discussed the popular paralleling of similar outputs with small value load balancing resistors, an example of a lossey combiner which also gives high loading interaction of the 2 outputs - increasing power but only indirectly improving the amp thru greater current output capability and effective decrease in load seen by each of the individual amps with shared outputs
I was pointing to the problems with some of the combiner proposals that keep popping up when you try to “improve” the output using different type amps - particularly the slow high power amp (LM38xx and relatives in this discussion) and a fast low power amp combination
Last edited:
Great idea to improve a cheap integrated amp, but whatever
you do, you will be restrained by the poor current capability of
these chips, not counting the SPIKE like dreadful protection circuitries.
The LM498XX + discrete power devices seems a better route.
you do, you will be restrained by the poor current capability of
these chips, not counting the SPIKE like dreadful protection circuitries.
The LM498XX + discrete power devices seems a better route.
Couldn't find anything about the PGA2300 anywhere.
Thanks - this makes a lot of sense as you'd said (and I concur) that CMOS switches need to be used into a very high impedance load to get good linearity. Using them directly in the feedback loop of an opamp where the impedances are likely only a few kohms spells trouble. If its crappy opamps which are the problem, then the lower voltage operation parts have a somewhat better one - with 10MHz bandwidth. That's why I was wondering if anyone had compared the 15V- with the 5V- parts. It might help to understand which was the worst offender. It seems likely if they just brought out the input to the opamp on a separate pin they could have a nice-sounding part. Whatever possessed them to introduce the potential of 31dB of gain from a 1MHz opamp is beyond me. Oh, I remember - its just a copy of a Crystal part isn't it?😉
Hi,
Funny, Amplifiers like Quads 405 combine two amplifier outputs easily, even though they do NOT use feed forward.
Quite possibly not even an "ideal" application needs to eliminate all interaction.
Sure, but how much?
This depends on the exact application. I have been working with FFWD in commercial applications, so I am not at liberty to discuss the topologies, but with some inventiveness you would be surprised just WHAT can be done.
And to correct a LM3886 or LM3875 a PDIP Video Op-Amp can work quite well, as we need to deal with VERY SMALL errors only..
There is no need for "crossover". The Error Correction Amplifier can run full range, as can the "combiner". However, the whole systems input should be bandwidth limited to the power bandwidth of the main Amplifier, something that is easily assured with a 2nd order LPF on the amplifier input.
Emphatically NO!
Tolerance is a job for tight tolerance parts, which is pretty trivial nowadays.
Ciao T
Funny, Amplifiers like Quads 405 combine two amplifier outputs easily, even though they do NOT use feed forward.
Quite possibly not even an "ideal" application needs to eliminate all interaction.
Sure, but how much?
This depends on the exact application. I have been working with FFWD in commercial applications, so I am not at liberty to discuss the topologies, but with some inventiveness you would be surprised just WHAT can be done.
And to correct a LM3886 or LM3875 a PDIP Video Op-Amp can work quite well, as we need to deal with VERY SMALL errors only..
There is no need for "crossover". The Error Correction Amplifier can run full range, as can the "combiner". However, the whole systems input should be bandwidth limited to the power bandwidth of the main Amplifier, something that is easily assured with a 2nd order LPF on the amplifier input.
Emphatically NO!
Tolerance is a job for tight tolerance parts, which is pretty trivial nowadays.
Ciao T
Hi,
Sorry, is it 2310/2311? Anyway, they have several variants that differ in rated THD (minimally), PSU Voltages (appreciably) but sadly not in any substantial way in sound quality.
I did. I found the 15V part sounded not only marginally better in application, but best, it did not automatically clip when you used a high gain Phonostage in front (like the EAR834P). I used to modify Shanling Amplifiers that used the 5V Parts. A common complaint was that "loud" sources would overload the Volume Control chip.
After chucking out the extra OPA2604 Op-Amp's Shanling build around the PGA devices, putting in a 15V part and changing the supply regulators for this part from 7805/7905 to better parts it still sounded so spectaculary bad that I fitted switches to bypass this and use external preamps (which of course kind of voids the point of buying an integrated Amp).
Yup.
The entity has at times been called Lou Cipher.
😛
Well, even BB must turn out the occasional total complete Dud, otherwise they'd be found to not comply with Sturgeons Law...
Ciao T
Sorry, is it 2310/2311? Anyway, they have several variants that differ in rated THD (minimally), PSU Voltages (appreciably) but sadly not in any substantial way in sound quality.
I did. I found the 15V part sounded not only marginally better in application, but best, it did not automatically clip when you used a high gain Phonostage in front (like the EAR834P). I used to modify Shanling Amplifiers that used the 5V Parts. A common complaint was that "loud" sources would overload the Volume Control chip.
After chucking out the extra OPA2604 Op-Amp's Shanling build around the PGA devices, putting in a 15V part and changing the supply regulators for this part from 7805/7905 to better parts it still sounded so spectaculary bad that I fitted switches to bypass this and use external preamps (which of course kind of voids the point of buying an integrated Amp).
Yup.
The entity has at times been called Lou Cipher.
😛
Well, even BB must turn out the occasional total complete Dud, otherwise they'd be found to not comply with Sturgeons Law...
Ciao T
Hi,
Agreed. HOWEVER, I would suggest you add a pair of LSK389 J-Fets as buffer (run at Idss and inf neccesary trim remaining offset with a fixed resistor) after the Noble Pot (and try to get a 250K one) and perhaps follow this with a 600 Ohm 1:1 line output transformer.
Yes, it is now an active preamp, with a balanced output. This may for example be used to drive multiple balanced Gainclones...
Given the degree of subjectivity and system context in Audio, yes, I suspect so.
Yes. I find that Gainclones build up mostly like Naim Amp's (big PSU Transformers, Big PSU Caps, Regulated Powersupplies [a'la Pedja Rogic]) do sound like Naim Amp's low down (which I consider a good thing) but much better as you go up the frequency ranges.
That is possible. In the recent decade really cheap speakers have been getting surprisingly good. The combination of low cost chinese manufacturing with western acoustic know-how has achieved a lot in that sector.
LM3875 based, low capacitance supply, yes? These can have problems driving "difficult" Speakers. The MS S442 probably drops quite low in impedance and is quite reactive in nature.
If I remember rightly, the MS442 is tri-wirable? I actually mistook the Speakers you have for the later Performance series, these where true 3-Way systems in Ceramic monocoque enclosures...
Mordaunt Short Performance 880
Maybe time to build a multiway gainclone with dedicated channels per driver?
The LF Drivers clearly want "Naim" Style Amp's. Do you know which Model Naim Amp the original owner used? It is not too difficult to combine Naim principles with Gainclones...
Hmmm. Maybe time for a paradigm shift?
BTW, if you can find them, one of my favourite british speakers of all time is theSD-Acoustics SD-1:
FS: SD Acoustics SD1 speakers - ribbon hybrids - The Art of Sound Forum
These would work rather well with a digital crossover (modded) and a Multichannel Gainclone. They are not all that hard to approximate in DIY, Audax PR170M0 Midrange (doped by hand, I believe Madisound now has a doped version as standard) in open back box, ribbon tweeter (Magnetostat actually) and a 10" Boxed woofer. If all done active with a digital X-Over very easy to get working.
The high efficiency Midrange (and likely HF if you use a modern magnetostat instead of the old Foster/Fostex Magnetostat in the SD1) will give explosive dynamics in active operation...
With a near 100dB/W/m Midrange (designing the open back box right can bring up the range below 1KHz up by around 6dB) you could even use a nice Pass Zen Amp or the like for MF. I think using the ribbon or magnetostat will also work okay with a Zen or similar Amp.
The bass can use a Naim Style Gainclone.
We can discuss more another time...
Ciao T
Agreed. HOWEVER, I would suggest you add a pair of LSK389 J-Fets as buffer (run at Idss and inf neccesary trim remaining offset with a fixed resistor) after the Noble Pot (and try to get a 250K one) and perhaps follow this with a 600 Ohm 1:1 line output transformer.
Yes, it is now an active preamp, with a balanced output. This may for example be used to drive multiple balanced Gainclones...
Given the degree of subjectivity and system context in Audio, yes, I suspect so.
Yes. I find that Gainclones build up mostly like Naim Amp's (big PSU Transformers, Big PSU Caps, Regulated Powersupplies [a'la Pedja Rogic]) do sound like Naim Amp's low down (which I consider a good thing) but much better as you go up the frequency ranges.
That is possible. In the recent decade really cheap speakers have been getting surprisingly good. The combination of low cost chinese manufacturing with western acoustic know-how has achieved a lot in that sector.
LM3875 based, low capacitance supply, yes? These can have problems driving "difficult" Speakers. The MS S442 probably drops quite low in impedance and is quite reactive in nature.
If I remember rightly, the MS442 is tri-wirable? I actually mistook the Speakers you have for the later Performance series, these where true 3-Way systems in Ceramic monocoque enclosures...
Mordaunt Short Performance 880
Maybe time to build a multiway gainclone with dedicated channels per driver?
The LF Drivers clearly want "Naim" Style Amp's. Do you know which Model Naim Amp the original owner used? It is not too difficult to combine Naim principles with Gainclones...
Hmmm. Maybe time for a paradigm shift?
BTW, if you can find them, one of my favourite british speakers of all time is theSD-Acoustics SD-1:
FS: SD Acoustics SD1 speakers - ribbon hybrids - The Art of Sound Forum
These would work rather well with a digital crossover (modded) and a Multichannel Gainclone. They are not all that hard to approximate in DIY, Audax PR170M0 Midrange (doped by hand, I believe Madisound now has a doped version as standard) in open back box, ribbon tweeter (Magnetostat actually) and a 10" Boxed woofer. If all done active with a digital X-Over very easy to get working.
The high efficiency Midrange (and likely HF if you use a modern magnetostat instead of the old Foster/Fostex Magnetostat in the SD1) will give explosive dynamics in active operation...
With a near 100dB/W/m Midrange (designing the open back box right can bring up the range below 1KHz up by around 6dB) you could even use a nice Pass Zen Amp or the like for MF. I think using the ribbon or magnetostat will also work okay with a Zen or similar Amp.
The bass can use a Naim Style Gainclone.
We can discuss more another time...
Ciao T
Well, if you're going to throw transformers into the mix...
Have the JFET buffer drive a nice 1:10 step-up (such as the CineMag CMMI-10C), and that drive a unity-gain stable chip amp such as the OPA-549.
se
Hi,
Yes, that is a possibility. Sadly I found non of the unity gain stable Power OPA's I tried (includes the BB/TI offerings and the Nat Semi LM12) sounded anywhere close to a Nat Semi "Overture" Amp (LM3875, LM3886, LM4780).
While it is possible to operate the Overture Amp's in unity gain as well, they do not really sound much better that way than with gain...
My proposal was more aimed at a general purpose preamp.
Of course, a 1:2 (150Ohm:600Ohm) stepup would work quite well in this case...
Ciao T
Yes, that is a possibility. Sadly I found non of the unity gain stable Power OPA's I tried (includes the BB/TI offerings and the Nat Semi LM12) sounded anywhere close to a Nat Semi "Overture" Amp (LM3875, LM3886, LM4780).
While it is possible to operate the Overture Amp's in unity gain as well, they do not really sound much better that way than with gain...
My proposal was more aimed at a general purpose preamp.
Of course, a 1:2 (150Ohm:600Ohm) stepup would work quite well in this case...
Ciao T
Folks,
For those that need a schematic to get what we are talking, here goes...
Ciao T
PS, with this circuit you get around 300R output impedance (still quite low) and bandwidth, distortion etc. is down mainly to the pot and transformer.
For those that need a schematic to get what we are talking, here goes...
Ciao T
PS, with this circuit you get around 300R output impedance (still quite low) and bandwidth, distortion etc. is down mainly to the pot and transformer.
Attachments
For that, the CineMag CMOQ-4 would also work very well and costs less than the CMOQ-1.
Using the LSK389B and CMOQ-1 or CMOQ-4, output impedance should be less than half that.
Transconductance of the 389B is typically 20mS at 3mA. That gives you an output impedance from the buffer of about 50 ohms. The output trannies, wired 1:2, would add 80 ohms to that. So you're looking at around 130 ohms.
se
Steve,
Probably. No Datasheet, so cannot tell.
Not quite Steve.
The datasheets have no DCR ratings, but let's assume 20 Ohm per section.
On the primary side (150R nominal) we have now 10 Ohm DCR from the transformer windings and 50 Ohm from the buffer (by experience, it is usually a little higher) but lets run with these numbers, the whole impedance on the primary is 60 Ohm.
Our transformer is set to 1:2 stepup, so it will step up imedances by 1:4, so the 60 Ohm from the primary appear at the secondary as 240 Ohm. If the two secondary sections are 20 ohm each we get 40 ohm, which we need to add to our 240 Ohm, so we get 280 Ohm.
As I was not interested in a precise figure, I just threw the 300 Ohm in the ballpark.
If that transformer was used 1:1 of course, then we would have around 130 Ohm Output Impedance.
I would posit that either 300 Ohm, 280 Ohm or 130 ohm are sufficiently low to not loose sleep.
Ciao T
Probably. No Datasheet, so cannot tell.
Not quite Steve.
The datasheets have no DCR ratings, but let's assume 20 Ohm per section.
On the primary side (150R nominal) we have now 10 Ohm DCR from the transformer windings and 50 Ohm from the buffer (by experience, it is usually a little higher) but lets run with these numbers, the whole impedance on the primary is 60 Ohm.
Our transformer is set to 1:2 stepup, so it will step up imedances by 1:4, so the 60 Ohm from the primary appear at the secondary as 240 Ohm. If the two secondary sections are 20 ohm each we get 40 ohm, which we need to add to our 240 Ohm, so we get 280 Ohm.
As I was not interested in a precise figure, I just threw the 300 Ohm in the ballpark.
If that transformer was used 1:1 of course, then we would have around 130 Ohm Output Impedance.
I would posit that either 300 Ohm, 280 Ohm or 130 ohm are sufficiently low to not loose sleep.
Ciao T
Not exactly point by point but some replies to Thorsten’s comments:
Quad uses a lossey combiner with R and L
I’m less familiar with "true error feedforward" (since many schemes reduce to negative feedback, with possible command feedforward) but it is easy to find Vanderkooy and Lipschitz’ analysis on the web and a quick skim reveals their claim that the Quad is a combination of error feedforward with interaction from the non-conjugate combiner that causes negative feedback stability issues
I think their points also apply to Stochino’s transformer series V source adder but I would need some time and motivation to tease out the details
It may well be that my comments about power requirements don’t apply to the best practical approximations of “full bandwidth” feedforward error correction – although I would warn that the attractive simplified heuristic ~ “0.01% THD 100W power amp only requires 0.01% * 100 = 10 mW rated correction amp” is wrong – IMD components with complex waveforms will have much higher distortion power in IMD components than the in the “simple harmonics” projected from a THD #
I find precision at better than 0.1% level difficult to deliver in a finished product – how about after a few years of aging, operating thermal cycling, environmental stress, ect. caused drift
LM38xx use large, presumably diffused Si, emitter resistors that would be expected to give several 10s % of GBW modulation over the >50 C die temp swings expected with Class AB operation – and excess loop gain will be < 100 over ~4KHz, giving > 0.1% thermal gain modulation regardless of feedback resistor quality at higher frequencies
Tight tolerance is a job for very high loop gain And high quality feedback parts
You may be constrained from showing current commercial schematics but most manufacturers will be printing the patent #s on the amp case or in the manual so disclosing the already issued patents describing these feedforward techniques to us wouldn’t seem to be asking too much
Quad uses a lossey combiner with R and L
I’m less familiar with "true error feedforward" (since many schemes reduce to negative feedback, with possible command feedforward) but it is easy to find Vanderkooy and Lipschitz’ analysis on the web and a quick skim reveals their claim that the Quad is a combination of error feedforward with interaction from the non-conjugate combiner that causes negative feedback stability issues
I think their points also apply to Stochino’s transformer series V source adder but I would need some time and motivation to tease out the details
It may well be that my comments about power requirements don’t apply to the best practical approximations of “full bandwidth” feedforward error correction – although I would warn that the attractive simplified heuristic ~ “0.01% THD 100W power amp only requires 0.01% * 100 = 10 mW rated correction amp” is wrong – IMD components with complex waveforms will have much higher distortion power in IMD components than the in the “simple harmonics” projected from a THD #
I find precision at better than 0.1% level difficult to deliver in a finished product – how about after a few years of aging, operating thermal cycling, environmental stress, ect. caused drift
LM38xx use large, presumably diffused Si, emitter resistors that would be expected to give several 10s % of GBW modulation over the >50 C die temp swings expected with Class AB operation – and excess loop gain will be < 100 over ~4KHz, giving > 0.1% thermal gain modulation regardless of feedback resistor quality at higher frequencies
Tight tolerance is a job for very high loop gain And high quality feedback parts
You may be constrained from showing current commercial schematics but most manufacturers will be printing the patent #s on the amp case or in the manual so disclosing the already issued patents describing these feedforward techniques to us wouldn’t seem to be asking too much
Last edited:
Hi,
Yes. But my point is that there is no need to have a non-lossy combiner.
Well, Stocchino uses two separate amplifiers that use individual feedback loops. One Amplifier (the super-fast Class A one) amplifies only the error of the second (slow, high power class B one) and injects it in the correct amplitude and phase after the output of the second amplifier.
Once the two signals are added (literally, by being inserted in series) the sum has cancelled the error of the second amplifier to that of the first amplifier.
Other than the local feedback loops no feedback is involved. The Feed-forward path is in no way bandwidth limited.
Sure, but if we are concerned about amplifier for music having just the first 10 % of the amplifiers power range compensated would already help.
And more crucially, lets us look what happens if the error amplifier in a feed-forward scheme runs out of available power? We simply return to the basic performance of the Main Amplifier. If the error amplifier is really speedy, it will recover speedily once the overload situation is past.
And each amplifier is fundamentally stable in itself and any interaction are minimal.
I buy 0.1% resistors routinely, they are not expensive, 0.01% are available. And most are very good in long term stability.
Ciao T
Yes. But my point is that there is no need to have a non-lossy combiner.
Well, Stocchino uses two separate amplifiers that use individual feedback loops. One Amplifier (the super-fast Class A one) amplifies only the error of the second (slow, high power class B one) and injects it in the correct amplitude and phase after the output of the second amplifier.
Once the two signals are added (literally, by being inserted in series) the sum has cancelled the error of the second amplifier to that of the first amplifier.
Other than the local feedback loops no feedback is involved. The Feed-forward path is in no way bandwidth limited.
Sure, but if we are concerned about amplifier for music having just the first 10 % of the amplifiers power range compensated would already help.
And more crucially, lets us look what happens if the error amplifier in a feed-forward scheme runs out of available power? We simply return to the basic performance of the Main Amplifier. If the error amplifier is really speedy, it will recover speedily once the overload situation is past.
And each amplifier is fundamentally stable in itself and any interaction are minimal.
I buy 0.1% resistors routinely, they are not expensive, 0.01% are available. And most are very good in long term stability.
Ciao T
I like the idea of a composite Class A / Class B amp. I'm guessing it would be more difficult than a GC, but definitely sounds very interesting. It's really only sensible to use this configuration with Midrange and HF channels, am I right?ThorstenL said:
As for the MS 442 speakers - I was actually very unfair towards them, as I first auditioned them casually positioned "somewhere" near the other speakers. After re-positioning them carefully, it turned out they are much more refined and detailed than Wharfedales, with tighter but weighty bass. I have now accepted them as my main speakers. My only problem with them is the midrange is slightly coloured. I can especially hear it in pianos and female vocals. It's not a huge problem, but I know it will keep bugging me until I replace them. I guess the reason why Wharfedales stayed with me for so long is that - true to their "Middle Field Monitors" name - they are completely neutral.
Also, I have replaced AD826's in my Marantz CD-52 with dual LT1028's, and I'm very pleased with them. AD826 were an improvement over the original opamps, especially in clarity and timbre, but they lacked sharpness badly; everything was very "rounded", and bass was boomy and bloated. LT1028 sound slightly dry in comparison, but this is because they are simply more accurate, and very quick, very neutral. To my great contentment, the "startling factor" has returned. Mordaunt-Shorts sound much less coloured with these opamps.
Dear Leon,
A semi discrete class-A amp doesn't have to be to complicated, if you use the LME49810/11 drivers from National. All the complicated stuff is already done (designing a good VAS) all you have to do is add the transistors of your choice, and bias. There are plenty of design examples out there with those chips. I can tell you from own experience the LME49810 with the STD03's darlingtons sound sublime, and easy to understand and to built. If you like, I have schematics of a working design.
With kind regards,
Bas
A semi discrete class-A amp doesn't have to be to complicated, if you use the LME49810/11 drivers from National. All the complicated stuff is already done (designing a good VAS) all you have to do is add the transistors of your choice, and bias. There are plenty of design examples out there with those chips. I can tell you from own experience the LME49810 with the STD03's darlingtons sound sublime, and easy to understand and to built. If you like, I have schematics of a working design.
With kind regards,
Bas
- Status
- Not open for further replies.