| pro |
Hi All.
I am building a 3 way active crossover with double op-amp (NE5532).
A silly question.
Because I use 1/2 op amp af filter and the other 1/2 as output buffer (voltage follower), if I want bias in class A the op-amp with a resistor or a CCS, witch one must I bias: the first one, the buffer or both? (or nothing).
Thanks a lot to averybody.
Ciao. |
|
|
| Upupa Epops |
| By this connection you get higher distortion ( " musical sound " ) and manufacturer can low distortion :cool: . Any manufacturer don't claim, which distortion is " nice ". |
|
|
| jacco vermeulen |
I suppose the reason for wishing to force the opamp in Class A is for improving the sound of the signal.
To me it would make no difference whether one of both amps in the 5532 acts as a buffer stage, as the signal passes both forcing both in class A would be the logical choice.
Personally i would go for a 634 combined with an opamp with a much higher slewrate than the 5532.
I admit it adds considerable to the cost with a 3-way filter but compared to overall cost the relative increase in cost is justifiable. |
|
|
| uli |
| quote: | Originally posted by pro
Hi All.
I am building a 3 way active crossover with double op-amp (NE5532).
A silly question.
Because I use 1/2 op amp af filter and the other 1/2 as output buffer (voltage follower), if I want bias in class A the op-amp with a resistor or a CCS, witch one must I bias: the first one, the buffer or both? (or nothing).
Thanks a lot to averybody.
Ciao. |
Just try and listen.
A good starting point whould be about 2k2 - 4k7 to - supply both outputs. The exact value depends on signal current drawn from output. You choose that resistor that the current thru it is at least equal to signal current calculated from load and Vout max.
Uli
:nod: :nod: :nod:
PS: Your only criteria should be your ears!!! |
|
|
| pro |
Do you mean that biasing in A class an op-amp, it seems to sound better because of the increased distortions? So it is better to don't do that.
Ciao. |
|
|
| uli |
| quote: | Originally posted by jacco vermeulen
Personally i would go for a 634 combined with an opamp with a much higher slewrate than the 5532.
|
634 is a good choice at output. I recommend swapping the aged NE5532 with OPA2134
Uli
:nod: :nod: :nod: |
|
|
| Upupa Epops |
| Make it like Uli wrote, uprising of sound will be surpraised for you ;) , Uli is correct :cool: . |
|
|
| uli |
| quote: | Originally posted by pro
Do you mean that biasing in A class an op-amp, it seems to sound better because of the increased distortions? So it is better to don't do that.
Ciao. |
Biasing class A changes the kind of distortion.
Look at peranders site -> the GC with class A biasing.
Go to www.passdiy.com and read NPs articles about biasing single ended class A which is the same as biasing an opamp with a resistor (CCS) at the output.
Uli
:nod: :nod: :nod: |
|
|
| Upupa Epops |
| Or build discrete opamp - example you can find on Bryston's pages ( they are using only eight cheap transistors ) :cool: . |
|
|
| pro |
Well, I am gonna using a 2134 for the high section, two M5238P for the mid section and a 5532 (the slowest) for bass section.
Now, if the filter drives the buffer, it means that looks at an input inpedence of the buffer very high. Isn't it 2k2 to 4k7 to high current for that (at 12V rail it's 5.4 and 2.7 mA), and the buffer with those values, must drive a 2k2 hom in parallel with 34K of the amp, that is about 2K. Isn't it a value a bit low for an Op-amp?
Ciao. |
|
|
| uli |
| quote: | Originally posted by pro
Isn't it a value a bit low for an Op-amp?
Ciao. |
you arent screwed to 2k2, just use the "appropriate" value.
This is about 50% of the load.
If you got a load of say 10k, a 6k8 for bias would be sufficient.
Better is a ccs . 10k load full swing about 15V peak -> 1,5mA.
build a 2mA ccs and youre done.
Uli
:nod: :nod: :nod: |
|
|
| peranders |
| quote: | Originally posted by Upupa Epops
Or build discrete opamp - example you can find on Bryston's pages ( they are using only eight cheap transistors ) :cool: . |
:cop:
Pavel, I strongly advise you to try to be more helping (if you really want to help that is) not telling "pro" to do something totally different.
We have seen pretty many posts of yours lately where you are to no help at all. So please Pavel.....
:cop: |
|
|
| peranders |
| quote: | Originally posted by pro
Hi All.
I am building a 3 way active crossover with double op-amp (NE5532).
A silly question.
Because I use 1/2 op amp af filter and the other 1/2 as output buffer (voltage follower), if I want bias in class A the op-amp with a resistor or a CCS, witch one must I bias: the first one, the buffer or both? (or nothing).
Thanks a lot to averybody.
Ciao. |
Pro, it's normal to have a class A circuit between an opamp output and a buffer input but how much good this circuit does is hard to know if you don't know how the output stage is designed in the opamp. It's not certain at all that you really will get an improvement, just by adding a circuit not knowing if you should tie it to V+ or V- or how much current you should have.
My advise here is that you start with no class A circuit to begin with.
From what I have seen not very many have bothered to used this class A ciircuit on this rather (or very!) old opamp. I have mostly seen it together with an OPA627, AD8610 or LT1115.
Notice also that it's not suitable to have this circuit if you have a normal load. It's best suited if you have a high-ohmish buffer as load.
Currents for the current generators are normally 1-2 mA, not more.
Inspiration can be found here:
http://tangentsoft.net/audio/meta42/misc/schematic6.pdf
http://www.sjostromaudio.com/hifi_f...r0schema_p1.pdf
http://www.linear.com/prod/datasheet.html?datasheet=202
http://www.sjostromaudio.com/hifi_f...r0schema_p1.pdf |
|
|
| pro |
OK. Thank you to all for the advices.
I think i'll start without class A, and then I'll arrange a small module where insert a j-fet or bjt CCS with the op-amp, inset it in the op-amp socket , maybe only for high and midrange section, and then listen to the difference.
I didn't find the Class A GC in the Peranders site. Because I am interested to it, could somebady give me the right link?
Thanks again. |
|
|
| uli |
look at his QRV4 project
Uli
:nod: :nod: :nod: |
|
|
| peranders |
| quote: | Originally posted by pro
I didn't find the Class A GC in the Peranders site. Because I am interested to it, could somebady give me the right link?
|
I'm afraid there aren't any such Gainclones :whazzat:
I can say so much that I haven't been able to pick any difference with and without this class A circuit but others can I'm sure.
You have the lnks above in those two project I have used this class A circuit. You may notice that Tangent has used a cascoded JFET, slightly better than a single transistor but you will loose some voltage. |
|
|
| classd4sure |
Hi,
Yeap, I'm going to get burned for this one!! lol..
I've heard of certain respected engineers rave about doing this to a 741 for audio purposes, I guess if you have nothing else around. So now I'm thinking for anyone who's really curious about this, it might be neat to experiment first with a totally abhorrent op amp like a 741, should be able to tell real fast what kind of difference it makes.
Though I of course agree for your actual circuit, do try it without first, it's not a hard addition to make later.
Also, I have it on good authority you should tie your CCS, or resistor, to the power rail which displays the least PSRR according to the data sheet, and that all it really does to improve things is bypass this 20dB worse PSRR, but then again I'd think since it is working class A the improvement should be two fold.
Keep an eye on the temperatures too.
Regards |
|
|
| uli |
| quote: | Originally posted by classd4sure
Also, I have it on good authority you should tie your CCS, or resistor, to the power rail which displays the least PSRR according to the data sheet, and that all it really does to improve things is bypass this 20dB worse PSRR, but then again I'd think since it is working class A the improvement should be two fold.
Keep an eye on the temperatures too.
Regards |
Depends on the design of the outputstage of the op.
As normally OPs arent complementary in design for best result you should take a look into the datasheet. Normally there is a circuit shown.
Uli
:nod: :nod: :nod: |
|
|
| richie00boy |
| I was under the impression that you need to connect it to move current from the output stage device that is in the emitter follower mode, in cases of quasi-complementary design. Is this right? e.g. in a classical NPN output stage the top transistor will be EF and the bottom CFP. You connect from output to -ve rail. |
|
|
| classd4sure |
| quote: | Originally posted by uli
Depends on the design of the outputstage of the op.
As normally OPs arent complementary in design for best result you should take a look into the datasheet. Normally there is a circuit shown.
Uli
:nod: :nod: :nod: |
Hi,
To what end?
I'm just guessing you would do that with the intention of using the NPN in class A and biasing off the PNP, however it doesn't seem to make sense if the NPN has -20dB less PSRR. |
|
|
| peranders |
| Which side that is the best is determined by the whole design, not just the output stage. |
|
|
| uli |
| quote: | Originally posted by classd4sure
Hi,
To what end?
I'm just guessing you would do that with the intention of using the NPN in class A and biasing off the PNP, however it doesn't seem to make sense if the NPN has -20dB less PSRR. |
You should bias the follower stage to class A, whatever kind of transistor it is.
If you have 20dB difference in PSRR, you should work on that problem first although implementing a ccs would be a good solution for that problem too.
Uli
:nod: :nod: :nod: |
|
|
| classd4sure |
| quote: | Originally posted by uli
You should bias the follower stage to class A, whatever kind of transistor it is.
If you have 20dB difference in PSRR, you should work on that problem first although implementing a ccs would be a good solution for that problem too.
Uli
:nod: :nod: :nod: |
Fair enough, but you say "..if you have..work on that problem first" as though it's unusual to have -20dB on one rail, from what I've seen, it's almost a given for any op amp.
CCS all around? |
|
|
| uli |
psu is an issue in the circuit I would work on that.
For example the OP2134 (which I like very much) has no different specs for plus and minus rails.
Uli
:nod: :nod: :nod: |
|
|
| classd4sure |
Good thread for sure but it was more of a debate on if it was worth doing or not, "just tie a resistor to the negative rail", this thread has already gone into advanced detail about how to do it properly. :D |
|
|
| peranders |
| quote: | Originally posted by uli
psu is an issue in the circuit I would work on that.
For example the OP2134 (which I like very much) has no different specs for plus and minus rails.
Uli
:nod: :nod: :nod: |
uli, I'm afraid the difference is 30dB! between V- and V+..... but what does these figures mean if your power supply voltage has a good quality? I think the whole debate forgot about this? I think we never got a real answer on this issue. I think also that this class A is a bit overrated and is a typcial internet rumour more than an important circuit. |
|
|
| classd4sure |
Good point.
If you took the measures of a very ..... stiff? /clean supply, then I guess you could just concentrate on the best output device to use, and EF would make sense, but in the case of complimentary, NPN.
Never any simple answers now are there?
Is it overrated? Good question, maybe not for a 741 :) Maybe for a high quality audio op amp. |
|
|
| uli |
| quote: | Originally posted by classd4sure
It does if you're looking in the right place, to my shock even the uncompensated opa637 has this symptom.
http://focus.ti.com/lit/ds/symlink/opa2134.pdf
page 5 "POWER SUPPLY AND COMMON-MODE REJECTION
vs FREQUENCY" |
oops!
youre right!
| quote: | Originally posted by peranders
uli, I'm afraid the difference is 30dB! between V- and V+..... but what does these figures mean if your power supply voltage has a good quality? I think the whole debate forgot about this? I think we never got a real answer on this issue. I think also that this class A is a bit overrated and is a typcial internet rumour more than an important circuit. |
Again I agree, biasing class A is just icing the cake when everything else is perfect (->PSRR etc.)
Uli
:nod: :nod: :nod: |
|
|
| peranders |
| quote: | Originally posted by classd4sure
Is it overrated? Good question, maybe not for a 741 :) Maybe for a high quality audio op amp. |
Have you tested a LM324/358? There you can speak about possible improvements! => extremly high crossover distortion! |
|
|
| classd4sure |
| quote: | Originally posted by peranders
Have you tested a LM324/358? There you can speak about possible improvements! => extremly high crossover distortion! |
Hi,
I sure haven't but that's great to know.
How about for a NE5532 such as the OP is using? |
|
|
| jacco vermeulen |
I have used both the 5532, 5534, OP27 , OP37, OPA627, OPA637, and a number of AD***/LT**** opamps.
Both in standard topology as forced in class A.
The 5532 may be an Oldy, but for the money it is still a good opamp, same goes for 5534.
Personally i like the Burr Brown sound, BB in all types sound way different from other manufacturers.
A PMI OP27 differs in sound from a BB OP27, that i noticed some 15 years ago when i started with PMI types on a pre, blew them when fiddling with the power regulation and replaced them by BB types, as my DIY friend did from first go.
Powersupply definitely has an effect, as it always has.
I have used OPA627's forced in class A on a pre with a battery powersupply, as with any class A amplifier circuit differences in the powersupply make a difference in sound.
But on any amplifier, AB or A, Grey Rollins wrote some lines on it on another thread stating the same.
I rebuilt two 4 channel class A car amplifiers i use in my Van.
The original KIA4558s i first replaced by OPA2134's, the TIP transistors by Sanken 1216/2922(16 of them).
Later i made a good deal on a hundred OPA627/OPA637's, placed a conversion PCB above the Dual opamp pads in able to use single opamps on the amplifiers.
As car amplifiers make use of switched powersupplies(which i am not a big fan of) the original car battery i replaced by two batteries with higher Amp*hour ratings, fitted under the truck.
On the battery lines are two 1Farad caps, internally a great number of 1000uF/ 4700uF EYF Roedersteins were added, the toroids swapped for larger ones, the switching transistors beefed up.
Both the 2134 and the OPA637 i later used were forced in class A, the sound benefitted much by any powersupply improvement.
Class A car amplifiers are sold as such, in reality the class A percentage is not that impressive.
The amplifiers in my van have been thermally coupled to the roof of the car, the wind over the railing on the raised roof actually cools the amplifiers.
Because of it i was able to raise the class A level, to an extent where sound quality suffered because the powersupply could not handle the load, so the bias had to come down.
My conclusion:
The Burr Brown OPA627 and OPA637 rank highest in integrated audiocircuits, even on switched powersupply car amplifiers.
Forcing opamps in class A the way Carlos described is an improvement, but it is not a wonderdrug.
Put a high level audio opamp in a **** amplifier, force it in class A, you will still have a **** amplifier.
Using an opamp such as the OPA627 also means paying extra attention to every other part of the amplifier.
Force the opamp in class A and even more attention for the entire layout is needed.
I started with a simple Opamp switch, and added a few resitors.
In the end i had to rebuild the entire amplifier, the entire powersupply, had to use several powerlines the size of my D..., even every powerline needed a separate Shaffner filter.
No work, no Play !:clown: |
|
|
|
