Simple Symetrical Amplifier

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hehehe nobody likes high distortion especially when it is third order. The only objection I have with this circuit is the relatively higher 3rd order distortion as compared to 2nd order. I would prefer higher overall distortion but from 2nd order.

You didn't do something wrong I believe. Here is a trick:

The FB resistor (1K) is connecting the input transistor's emitor and the output.

You want the input transistor to have Vbe of at least 0.6v (approximation). Base voltage (Vb) is assumed zero and so is output voltage (Vout). So to get 0.6v at the emitters (Ve) you want a certain current thru the FB resistors (This is what the CCS is for).

V = 0.6V
R = 1K
I = calculate! ---> 0.6mA.

CCS current == current thru FB resistor + current thru transistor

Note that the current thru the transistor is also affected by the collector resistor (1K here, which is a good value). This current is around 0.5mA. So more or less you want the CCS to give out 0.6mA+0.5mA = 1.1mA.

You may want to set both CCS to output 1.2mA. You will find that the higher the current, the lower the distortion. That's why you got ("too") low distortion when you use higher zener voltage which affect the base voltage of the CCS, giving very high current I believe.

But I suspected that it would be better (e.g. for the stability) if top CCS gives different current than the bottom CCS. With positive voltage at the output you want the top CCS to have slightly higher current. You can try 1.2mA for the top and 1.1mA for the bottom CCS.

That's just my guesswork, you will have to troubleshoot the correct setting after you build it (if you intent to). In Christophe's circuit, top CCS has 3K3 and bottom CCS has 3K32. This means that the top CCS has slightly more current (I guess). But unfortunately the base voltage of the ccs also determines the current. I guess that you need to work out the 3K3 resistors on the CCS and the 30R resistor at the base of the VAS transistor.

You owe to yourself to build any of the SSA amp. Better is the more complex one.

I know how to change those things. Problem is that I whanted to repeat Esperado's simulation, and my result are catastrochic(with 3.9V of a voltage source instead of 3.9V zener). Bandwidth is up to 14kHz and distortion is terible.
What should be the VAS current? It seams that with your suggested CCS current, it is to low for my liking(1-2mA). JLH used 10mA in his 80W MOSFET where VAS drives laterals directly.
dado
 
By the way, the most elegant way to set CCS current is changing D1 and D3. In this case the cascode voltage is changed in the same time, and in this case is incremented for better work of Q7 and Q9. With a zener of 5.6V CCS current is set to 1.5mA and VAS current to 14mA. Phase marigin was not changed(47degree). Distortion drops even more.
dado
 
I know how to change those things. Problem is that I whanted to repeat Esperado's simulation, and my result are catastrochic(with 3.9V of a voltage source instead of 3.9V zener). Bandwidth is up to 14kHz and distortion is terible.

Have you simmed the other SSA circuits? They are all terrible. But it has been reported that measured performance is better than simulation.

This SSA front end is intriguing.

What should be the VAS current? It seams that with your suggested CCS current, it is to low for my liking(1-2mA). JLH used 10mA in his 80W MOSFET where VAS drives laterals directly.
dado

The CCS is for the front end, not VAS. You have to pay attention to the operation of the input transistors. But you're right, that the VAS current is too low (tho it is not the determinant factor) and I don't like most circuit where output is driven by VAS. That's one reason why I have never built this circuit (all components, parts, everything is already prepared, just need soldering).

I have tried to design some "better" circuits based on this SSA_Crescendo, with hot VAS transistor etc. The only thing I couldn't achieve was stability! But I suspect that by "trimming" the CCS (e.g. top CCS has more current than bottom CCS) stability can be achieved better, but I haven't tried it.

By the way, the most elegant way to set CCS current is changing D1 and D3. In this case the cascode voltage is changed in the same time, and in this case is incremented for better work of Q7 and Q9. With a zener of 5.6V CCS current is set to 1.5mA and VAS current to 14mA. Phase marigin was not changed(47degree). Distortion drops even more.

Actually there are many better ways when stability is not an issue (e.g. separated ccs or bias networks, use of voltage dividers, etc).

Problem with zener is we cannot trim a zener voltage easily, unless we put a resistor in series like Christophe did. It is easier when all voltage is controlled by "voltage divider networks". It is easy to achieve good performance but stability. And this SSA circuit is tricky. We must have precise model and measured transistors, which we don't. I just don't want to risk my laterals during tweaking, that's all. But I would love to see you continuing your simulation to building. At least, you are more experienced than me.
 
Have you simmed the other SSA circuits? They are all terrible. But it has been reported that measured performance is better than simulation.

This SSA front end is intriguing.



The CCS is for the front end, not VAS. You have to pay attention to the operation of the input transistors. But you're right, that the VAS current is too low (tho it is not the determinant factor) and I don't like most circuit where output is driven by VAS. That's one reason why I have never built this circuit (all components, parts, everything is already prepared, just need soldering).

I simed a time ago some of the other SSA circuit with no good result so I left it be.
Now, again, I tried with this from Esperado and my result a far away from Esperado's even if he does not say, for THD, at what frequency and power sim was done.
I would like to see his FFT also
Yes CCS is for front end, but changing the CCS current the VAS current was changed too.
For the moment I do not have intention to build SSA(all my laterals are in JLH MOSFET amps) as I am occupied to finish my six channel TT amps.
dado
 
I simed a time ago some of the other SSA circuit with no good result so I left it be.

You have to know where those bad numbers came from in order to get a good SSA. The input transistors are the key.

Now, again, I tried with this from Esperado and my result a far away from Esperado's even if he does not say, for THD, at what frequency and power sim was done.

If you properly choose the front end transistors, I believe you will not see bad distortion in real life as you see in the simulation. To "see" (or predict) the performance of this circuit you can replace the feedback caps with 0.6V of voltage source. In real life the voltage across the FB caps can be higher or lower than 0.6V. If you get too low voltage then you will get terrible simulated distortion which depends on the model used by the simulator to model the front end transistors.

As far as I remember, those numbers presented by Christophe are true for all listening level SPL (0.0015% at 0.3Vin, 0.002% at 1Vin or more), 1KHz.

I'm agree with you that the bandwidth is too high. Not that it is bad (actually it is good), but we can concentrate to improve other parameters and allow for lower bandwidth as the consequences.

Yes CCS is for front end, but changing the CCS current the VAS current was changed too.

Technically it doesn't have to be like that. There are many things that affect VAS current, independent of the CCS current to bias the front-end transistor.
 
Disabled Account
Joined 2006
Increasing the FB resistors (e.g doubling the resistance with 2K||75R) will also solve the issue.

But all of these will put the circuit in worse stability. This is probably the reason why this topology is not popular, despite its good sound character.

Nice interpretation, I suggested this weeks ago, no need to worry about the stability, it gets affected so little it is of no consequence.

There is no reason for SSA THD figures to be high, I showed my 10 year old design which is the same circuit as having 0.005 distortion at 20khz at 30 Vrms. There are about another 5 or so identical commercial designs using the identical (actually simpler design) all showing THD figures below 0.01.
 
Actually I meant stability against oscillation, the hallmark of all high speed circuits.
Well, if your amp has a flat banditch with no peak at the hight end (FB compensation), and no overshoot with square waves (additional input low pass) there will be no stability problem.

Ah, so it will save the speaker when there is oscillation.
Yes, my protection compare input and output signals after output level has been reduced in dividing-it by the amp gain factor. If any difference (DC, HF or even distortion) it will fire the protection, instant because there is no integration or DC filtering.
In fact, this protection is so sensible that it can fire at each Kick drum on a real loudspeaker (that gives us a good idea about how are working any amp in real world with the current generated byt the moving coil (i dont know how to translate 'force contre electromotrice'). So i have to tune a gap.
Unfortunately he didn't use lateral output. I guess the CSA will use the BIGBT output. :eek: But hopefully your amp will outperform the CSA so I don't need to build the CSA :rofl:
I'm not so sure of that. Listening comparison will tell-it. And, about stability issues, i believe SSA has been tested on several samples, my amp was just tuned as-it (i had it for years before the modification) with *my* old components.

As i said, i published it for educational purpose, and do not encourage people without enough electronic skill to build-it as it. Even the original Crescendo in his VFB original version presented stability issues for several DIYers, while mine was rock stable.
I first builded mine for real, and i made this sim several weeks after. I'm not even so sure components values are exactly the same, My amp in in my enclosures, and i'm too lazy to disassemble -it to look at the real values.
Nothing new here, Murphy said that amps always oscillate and oscillators never start.
 
Last edited:
Spice models of the lateral mosfets, which are part of the Esperado's simulation file are completely useless. The gate threshold volatage is to low, and I changed them for Cordell's 2SK134/2SJ49. Not sure for the rest of the models, but for now I used it.
Here is my simulation. I set for 14mA of the VAS current, incremented feedback resistors and it looks that is now more stable with Phase Marin of 60degree and Gain Margin of 20dB.
Decreased power voltage from +-70V to +-50V, I think that two pairs of those lateral are not enough if a load drops to low and +-70V is to much for this amp.
dado
 

Attachments

  • dado2-ac.jpg
    dado2-ac.jpg
    258.6 KB · Views: 623
  • dado2-distortion.jpg
    dado2-distortion.jpg
    240.4 KB · Views: 593
I think that two pairs of those lateral are not enough if a load drops to low and +-70V is to much for this amp.
I used-it with +-75V in his VFB version for years with 6 Ohms load. (i use to compensate my speakers both motional and zobel to get a near flat impedance on all the bandwitch.).
Well 3 // units on each side would be better to have a perfect value of the impedance hole of the power fets, so: more stable.
Again, i do not see any interest in trying to build this crescendo amp, for those who do not have it yet. Components are obsoletes (for some, it is a pity) printed circuit no more available, etc... The only interest, is to improve this amp for those who have-it yet, on my point ov view.
About sims, i'm of the old school. we do not had this tools, in our laboratories, when i was on the business. Just multimeters, volt meters, osciloscopes, distortiometers. and slide rule :hypno2:

About simulation values, as, on the VFB version, the spice models used give quite good result comparing to the real amp measurements, i do not think there are to far away from my parts.
And, on my CFB amp version, when i tuned-it, the square waves where ok for that kind of slew rate: Limit of my generator.
 
Last edited:
I used-it with +-75V in his VFB version for years with 6 Ohms load. (i use to compensate my speakers both motional and zobel to get a near flat impedance on all the bandwitch.).
Well 3 // units on each side would be better to have a perfect value of the impedance hole of the power fets, so: more stable.
Again, i do not see any interest in trying to build this crescendo amp, for those who do not have it yet. Components are obsoletes (for some, it is a pity) printed circuit no more available, etc... The only interest, is to improve this amp for those who have-it yet, on my point ov view.
About sims, i'm of the old school. we do not had this tools, in our laboratories, when i was on the business. Just multimeters, volt meters, osciloscopes, distortiometers. and slide rule :hypno2:

About simulation values, as, on the VFB version, the spice models used give quite good result comparing to the real amp measurements, i do not think there are to far away from my parts.
And, on my CFB amp version, when i tuned-it, the square waves where ok for that kind of slew rate: Limit of my generator.

I am old shool too(retired telecomunitation engineer) and Spice simulator is great tool, so I am trying to lern it now.
Simulation with your lateral transistor models shows low distortion too(when I changed CCS zener from 3.9V to 4.7V) but with wrong resistors values. One of them is R33 8ohm resitor, and with 1.13mA of the VAS current produces 138mA per output transistor pair. This impossible in real world, so the models are wrong.
dado
 
Finally got this thermaltrak version in stereo and housed.

http://www.diyaudio.com/forums/atta...metrical-amplifier-ssa-simple-thermaltrak.jpg

Not as neat as LC's fabulous versions, but quite functional. The SSA amps are on the vero board at the rear of the chassis. The other two amps are symasyms - for a four channel amp (still need a set of binding posts and rca jacks for the symasyms). Early listening to the SSA shows no defects that I can hear. Clean, clear music, powering the range above 300Hz so far. I like it.

Technical: Gain of 23dB. No noise at all with 100dB+ speaker. Very slight low frequency blip at switch on (100mV offset, maximum). Not audible with my speakers. No audible blip at shut off. Offset wanders around for a minute or so, but settles down to less than 20mV. Nice square wave at 100kHz - slight, well damped overshoot with a load, a little more pronounced without.

Practical: It takes a while to tune in the bias and offset, as they are interdependent. Once done, it's stable so far. If I were building again, I'd adjust it close, then measure the 500R pots on the current sources. Then I'd replace with a resistor and 50R pot, as it's quite sensitive to small changes and takes some fiddling to get within 10mV. But I'm not highly motivated to order the pots and change now, as I've already done the fiddling. The input transistors are from the same batch and well matched within type, but no matching across types. The driver and outputs are not particularly well matched. They are what I had.

Did some tweaking, as the bias drifted more than I liked. It seemed to be over compensated. As the overall amp warmed up, bias moved down and the spread between start-up and steady stage was about 2 fold. The steady state point tended to go lower when the amp output was increased. With the amp just on the heatsink, I didn't see this, as the input and drive transistors were exposed to room ambient. With the amp all cased up, they were in a warmer environment, and the NTC resistor overcompensated without the 220R (I think that was the original recommendation) series resistor. I tried 180, but that tended to undercompensate, so I settled on 130R in series with the NTC. While I was at it, I substituted 180R and a 50R pot for the 500R pot that sets the CCS current. This made adjusting a lot less touchy. End result is that bias starts out at about 5mV (0R1 emitter resistor) and climbs to about 12mV over a couple of minutes, then stabilizes at about 16mV a few minutes later, and stays there over hours. With a 5W sine wave output, the bias drifts up less than 2mV as the amp temperature goes up a little. That would more than cover any average power requirement that I would need.

I believe that the main issue for compensation is the fact that the input pair, drivers, and CCS, all contribute to the bias drift. The NTC provides a single compensation which must counteract the summed effect of the three contributors. In my case, they each reach equilibrium at different rates and reach different temperatures in response to a change in their ambient setting. If someone were to do a similar version, I would recommend trying to get the input, driver and CCS (bottom of the cascode pair anyway), on a single sink. Then they would all reach similar temperature together and compensation should be fairly easy to get right. The outputs, on the main sink, are well compensated independently.

Scope pictures look good. Only a little overshoot, visible here on a 100kHz square wave (5mV/div, 10x probe). Sine waves don't show it, except a couple mV spike on the peaks at 1mHz - not visible as amplitude is increased. I'm inclined to leave it as is.

Sheldon
 

Attachments

  • SSA 100kHz.jpg
    SSA 100kHz.jpg
    47 KB · Views: 1,211
Last edited:
Hello everybody!

Glad to see this great thread moving forward(which it should).

I'm here to share the story of my re-incarnated SSA which I built a week ago.

As the pictures show I built the front end and the VAS+cascode on different boards and interconnected them with 7 wires. BC550C/560C as input, same as cascodes, BD139/140 regulars as VAS, 1N4148s for bias setting and 1058/162 Laterals as outputs. The diodes set the bias to a nice 150mA through the FETs with +-25VDC supply. Total Iq is 180mA. 5pF silver mica caps as bandlimiters in the VAS and 5.6pF(Old Philips ceramic) as input RF filter. Offset control with two 10K small pots from +-15V zeners, filtered with 1uF siemens MKT. Power to the VAS and the front end is supplied through two 1N4007s with 1uF before them and 1uF+1000uF after. 100ohm gate stoppers and no Zobel at the output.

The amplifier worked like a charm the very first time it was turned on and after setting the offset I was listening to music within 5 minutes of the first run.

No hum, No noise, No offset problem, No overheating, No bias drifting, No nothing!!!!! What is this amplifier!?

Music didn't stop for two hours. And people, what dynamics, what purity, what bass, what mid and overall, what sparkling treble! The speakers were singing!

Honestly, I think my fullrange drivers are not truly full-range at all. But I didn't realize what they were capable of until they met SSA! I have found the holy grail of amplifiers!

Great vocals, both male and female, sitar, sarod, tanpura, mridangam, sarangi, guitar, piano, drums, flutes, violins, cellos, saxes, kicks, claps, cymbals, crashes, snares, bells, synthetics... EVERYTHING sounds JUST RIGHT... what a pleasure!

Tonal character unchanged from very low volume right upto very high! How does SSA do this? No other amp I have built or listened can successfully show this charecteristic and SSA does it with ease, even with my 4" $2 drivers! How, where's the magic?

The sound is absolutely non-disturbing for long listening trials. I live with my mom and sis. And they are excellent and sensitive natural scopes:)P) for the testing of this very parameter. In the past they have helped me accurately decide an amp's "obstructive" or "intrusive" or "artificial" or "just bad" character. They say that when I play music through a good amp it feels like someone else is present in my room and is singing or like there are real musical instruments being played in my room, feels real from outside. For the SSA, their remarks are "NATURAL" and "LOVELY".

All these with a 5 meter coax from an AC'97 sound card and an input cap which is, ahem, a cheap bipolar 10uF electrolytic.

SSA is a milestone for its simplicity, a benchmark for its speed, a crown jewell for its music.

My drums mentor, the man I respect the most told me that very good things will happen to me this year. One happened a just week ago.

LC, Thank You. We are lucky to have you around.
 

Attachments

  • DSC00132.JPG
    DSC00132.JPG
    479.1 KB · Views: 258
  • DSC00131.JPG
    DSC00131.JPG
    529.1 KB · Views: 262
  • DSC00129.JPG
    DSC00129.JPG
    504.9 KB · Views: 258
  • DSC00128.JPG
    DSC00128.JPG
    383.5 KB · Views: 760
  • DSC00127.JPG
    DSC00127.JPG
    317 KB · Views: 774
  • DSC00126.JPG
    DSC00126.JPG
    461.2 KB · Views: 890
  • DSC00125.JPG
    DSC00125.JPG
    475.6 KB · Views: 952
  • DSC00133.JPG
    DSC00133.JPG
    464.2 KB · Views: 223
  • DSC00134.JPG
    DSC00134.JPG
    509.5 KB · Views: 244
  • SSA.PNG
    SSA.PNG
    33.3 KB · Views: 599
Congratulations

LC, Thank You. We are lucky to have you around.

Hi Shaan, I am flattered, thanks, should rename myself to Lucky Cat ;)

Very very positive impression regarding music reproduction from you Shaan, Nico Ras, Bigun, Esperado, Sonnya, MiiB, Joachim Gerhard, Sheldon, Lazy Cat, Jazzz (SSA owner) and maybe somebody else in background. Still waiting Marc to finish TSSA and of course all others invited. :yes:

What surprises me the most, all the guys from above made each their own PCB-s, some even without and some PCB in P2P like you Shaan (pure work of art!) and in spite of all kinds of good amps available still listening to SSA/TSSA. :up:

What can I say, I am honoured to know you all my friends, enjoy the Sunday and good music, regards Andrej :wave:
 
After two weeks of daily operation I checked the offset and bias today. Both unchanged. Whoa!

Probably the best fortnight spent with an amplifier. There has only been signal, no noise. So the psychological SNR or satisfaction is infinite mathematically. :up:

I have some questions. I did not feel a hint of distortion from the class-AB SSA. If the SSA with source follower config already sounds(and measures) so well, then why is there the drain follower TSSA with limited power, hot sinks and big Zout? For the fostexes and lowthers? Or is there a noticeable increase in audible and measured performances? Does it sound the same(or better) through the same speaker drivers? Is it an attempt to eliminate as many gain stages as possible? Or is it for demonstrating the scalability of the SSA front end?
 
Esperado is using CSS instead of cascode on the input stage, what is better ?
No way to answer to this kind of question: "What is better". As each circuit has his own sound/advantages/disadvantages.Current mirror brings an advantage on bandwitch: As the frequency of the first stage increase, his efficiency, so his current consumption increase. Then you have some sort of auto-compensation with an increase of the resultant bandwitch. That you will pay at the extreme hight end with a highest Q roll-off., IE phase rotation.
I would like made an amplifier with lateral mosfet for my Magnepan 1.7 (86db 4ohm)
What do you think about using SMPS sypply for SSA amp ?
Again, you'll have to listen, compare, to make your own subjective conclusions. Just remember that, if Current feedback topologies brings-you reduction of distortion as well as increased bandwitch, you will pay it by a worse ripple rejection of the power line. So SMPS will need great filtering care of it switching frequency. But you will win on price, compacity / efficiency, dissipation, regulation. I prefer SMPS for analog amplification, linear power supply on Class D amplification. Some will prefer the contrary ;-)
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.