5534 audio amp

klitgt

Member
2004-09-16 9:31 am
Denmark
Wonderful question!!

in DK there is a discussion going on regarding the bipolar NE5534 versus FET based opamps. BTW NE5532 is not a dual 5543. 5532 is unity gain stable, 5534 is stable at a gain of 3 or more + other minor differences.
If possible you should use 2 x NE5534A (one for each channel) and not one NE5532 (½ per channel) for audio. The A grade is specially made for audio.

Now to the war between people who love and those who hate NE5534.

Lovers say NE5534 is much more musical than FETs.
The opposite point of view is that FETs are less grainy, especially in the high octaves, and therefore "sound better".

My own opinion is based on the AD825 (FET) versus NE5534.
The AD825 is perhaps more easy to listen to. On the other hand the NE5534 is more dynamic with better punch in the bass and up in the mids.

But if you are willing to pay, the OPA627 is the best of 2 worlds.

It is still woth considering that one of the most rewarded brands of poweramps, the Gamut, uses NE5534A in the balanced input stage to convert it to unbalanced before the signal goes into the following stages of the amp.

Kind Regrds!
 
klitgt said:
Lovers say NE5534 is much more musical than FETs.
The opposite point of view is that FETs are less grainy, especially in the high octaves, and therefore "sound better".
If you really want to compare, each opamp type must work under optimal condtions and this isn't the case, very often. You don't really compare the opamp itself rather where it is.
 
5532's and 5534's hummm.

Well, these op amps have been the stable in most commercial equipment since the dawn of their existance. These have worked very well with minimum problems. Would their be an advantage to using something else? Looks like that would be more of a matter of personal taste or that hidden something that tells you that a piece that costs 10 times more is really worth the extra money.

To answer the question (Is 5534 still a good choice for audio amp (pre-amp) design?)

In my opinion, yes either the 5532 or 5534
 
I think they're best used for low (<10) inverting gain and low (<10k) source impedances, if you need high noninverting gain with high quality for phono preamps then OPA627/637 is worth the extra cost.

NE5534 and LM318 are the only op amps I know of that bring out the collector terminals of the 1st LTP stage (Pins 1,8), this lets you add a discrete 1st LTP stage without added phase shift, also the 5534 comp pin (pin 5) can be used with 5 - 10mA current source for a class A CFP output stage that bypasses the on-chip short-circuit protection, refer to this post, this post, and the schematic in this 5534 datasheet.
 

tlf9999

Disabled Account
2005-05-25 2:25 am
none
nuvistor said:
Ialso the 5534 comp pin (pin 5) can be used with 5 - 10mA current source for a class A CFP output stage that bypasses the on-chip short-circuit protection,

Why do you have to spill out all the secrets about this wonderful chip? :)

That has been the single biggest advantage (for me anyway) of this particular design. It saves you time and effort to design your own input+VAS.
 
Why do you have to spill out all the secrets about this wonderful chip?

I've known about the 1st stage trick for almost 25 years, but I have to admit that I didn't understand that the 5534 output stage was NPN totem pole and that you could sink significant current from pin 5 before last year, after poring over its schematic for probably the 50th time. Hopefully others can enjoy this knowledge. :santa3:

If you have noninverting gain you really should try the 1st stage trick, with cascoded matched dual JFETs if you can, or matched NPN with emitter resistors. IME almost as good as an OPA627 and much better than the 5534 for clarity, and the reduced OL gain allows unity gain without added Ccomp.
 
Several years ago I was quite amazed, that the input stage of that NE5534 has massive head room.
I measured this, when "discovering" TIM distorsions due to
overdrive events of the input stage. Under normal conditions
the input stage only uses a small percentage of it max. swing.
(If I remember right: I measured the signals pin 1 & 8 at 10kHz sine wave 10V out put and found that it used only 20% of its clipping limit.
So it really needs bad HF steps in the audio signal to overwhelm this
old timer OpAmp).
I found this quite amazing, because you get principally more noise the more head room you spend there... and noise of that beast is also fine!
I tried to compare this with the higher graded OP37 that times and
TL072, but Ne5534 seemed to be outstanding.
Also my ears found them OK.

...would be interesting to compare them with some more
modern OpAmps like AD8610 / AD8620....
 
Interesting comments on input stage headroom. You don't mention feedback network configuration, assuming unity-gain follower with 22pF Ccomp and 10kHz input, 5534 has Avol of 60dB, input stage gain is 13k/(0.026/200uA) or 100, so 2nd + 3rd stage gain is 10, and voltage on pins 1,8 would be 10V/10 or 1V. IOW the negative feedback reduces the 1st stage voltage swing by the available Avol and the closed loop gain Acl, more Avol and less Acl will reduce the swing.
 
nuvistor said:
Interesting comments on input stage headroom. You don't mention feedback network configuration, assuming unity-gain follower with 22pF Ccomp and 10kHz input, 5534 has Avol of 60dB, input stage gain is 13k/(0.026/200uA) or 100, so 2nd + 3rd stage gain is 10, and voltage on pins 1,8 would be 10V/10 or 1V. IOW the negative feedback reduces the 1st stage voltage swing by the available Avol and the closed loop gain Acl, more Avol and less Acl will reduce the swing.

...urghs, should have documented that, but I didn't.
The set up was a higher gain non inverting amp (Feedback 2k2 and
100 Ohms from inv input to ground.... or something like that....)
I used no Ccomp.
From my understanding the signal on pin 1 & 8 at stable 10khz sinewave is determined by the output swing and the gains between output and collectors of the input stage.
Some of these gains have f3-corners much below 10kHz, so with increasing frequency the input stage will have to deliver more signal to achieve a constant magnitude at the output of the OpAmp....
For rectangular input signals you will get massive peaks at the input and collectors of the input stage as the OpAmp output will slope slower and less signal feed back. Means higher error signal and with this massive regulation action. The peaks at the collectors of the input stage are not uncontrolled overshoots, it's a proper error correction. .... well, it is proper as long as the input stage does not clip...