New Doug Self pre-amp design...

Folks,

After much off topic stuff discussing literary accomplishments and personal integrity etc., lets go back to the topic for a moment.

I took exception to the need for a very large number of operational amplifiers in order to accomplish a linestage with 16dB gain, tone control and low noise/distortion. I proposed an alternative option as of below.

I would suggest for anyone interested in tone controls, active line stage and simple circuits with high quality parts and no need for any more than one good OPA and one (optional) simple discrete buffer may wish to look here (see earlier posts for the single transistor plus LM317 Buffer):

http://www.diyaudio.com/forums/analog-line-level/194119-high-end-tone-control.html#post2688984

237641d1314541878-high-end-tone-control-active-baxandall-sch.gif


This circuit can be fed by any volume control we care to use, however a 10K control should be considered if the circuits noise is not to be increased unduly. The tonecontrol has a fully transparent bypass option and can in principle have selectable corner frequencies (not shown)

As shown the linestage has around 16dB gain and -100dBU unweighed output SNR, worst case, 10K volume control.

When driving a 600 ohm Load to +19dBU the UNBUFFERED linestage will produce 0.003% THD(20K) and 0.0002% THD(1k). Adding a suitable buffer can extend this performance to lower impedance loads and can reduce the THD for 600 Ohm loads.

Without the addition of any RC Filters the -3dB Bandwidth will be DC to 2MHz.

(all numbers based on OPA627 Datasheet and Johnson noise of the circuit impedances)

Now everyone was eager to point out my many shortcoming, but only one technical criticism was presented, namely that in my suggested circuit turning up the treble control would present a severe load for the circuit and would thus drive up THD appreciably.

Now I am waiting still for someone to post numbers for Self's circuit (which incidentally also present an Op-Amp somewhere with a VERY LOW load impedance at HF and thus will drive up THD) under identical conditions and as remarked, the circuit I suggest can be fitted with a very simple buffer which would eliminate the grounds of this criticism, could it be shown to be neccesary to match D Selfs deign.

So let us add the single BJT and LM317 Class A buffer both for street cred and to shut up the muttering noise from the cheap rows.

To extend my tone control circuit to a full preamplifier I would suggest we need to add a volume control (10K Audio Taper) and a Balance control. As the traditional balance controls used in the 80's which had zero attenuation in the centre position seem hard to get I would probably just use a 10K linear pot connected as series resistor before the volume control.

If we are happy with around 10K source impedance our pre-amplifier is finished, if not we may need to add an input buffer to offer a higher input impedance. In the simplest solution this could be another OPA627 configured as follower and offering 1M Input impedance, which could be switched to balanced input duties for balanced inputs.

That said, my personal (budget busting) solution for a balanced line input would probably be a Jensen JT-10KB-D which also helps attenuating the silly full scale levels studio gear puts out to something sensible, directly into the volume control.

To add a balanced output we could just a third OPA627 with the same buffer as above, or a seperate driver circuit such as the BB DRV134, but my personal choice would fall on a Jensen JT-11-BMCF, another budget buster of course.

In terms of many measured parameters I would suggest what I propose will at least equal D. Selfs design in terms of measurements and may well exceed it, at much less complexity and depending on choices for a comparable or even lower budget.

The most ironic part perhaps is that what I have described is a "generic" 1980's japanese preamp, except for the OPA627 Op-Amp's, though some of the discrete circuits in such preamp's, like for example in the Luxman Advance Z-502 Kit perform quite comparable, except on noise due to their selection of a higher resistance Volume control (50...100K).

Ciao T
 
Hi,

Wow, I haven't seen a tone control like that since back in the days when people still used valves for audio! ;)

It has actually never fallen out of favour in Japan, it was still found in Japanese integrated Amp's in the oughties...

The beauty is that when defeated nothing is added in the signal circuit and the additions when operating are limited to the frequency response alterations... Plus, there is no variable loading of the source and the input impedance may be made as high as desirable, which helps reducing the distortion introduced in the track/wiper interface of the volume control, when there is a load on the wiper.

If we use a circuit with negative feedback this design is a very smart choice. I suggest you read the original thread this came from...

http://www.diyaudio.com/forums/analog-line-level/194119-high-end-tone-control.html

Ciao T
 
Hi,

No DS fan will volunteer that move, its already known to those who are skilled in art about what would be the outcome under full treble boost.

Pitty that.

Now what would be the consequence of turning the treble up to full in D. Selfs design?

Increased THD? Perish the thought.

Maybe if DS instead of paralleling cheap Op-Amp's used my Class A buffer this could be avoided... ;-)

Ciao T
 
Hi,



But that is not the circuit D Self uses and this circuit alone is not useful for a complete preamp (it needs a buffer before it) while mine is.

So, anyone for D Self's circuit at 7V RMS output and the treble control fully up?

Those are the values that Doug Self used.

I did add a buffer in two versions , inverting buffer so the whole circuit
is non inverting , and a non inverting buffer that will left the circuit
as an overall inverter.

Of course , the buffer is a NE5532.

I would point that the current requirements of doug self circuit
are overkill , i would increase the baxandall impedances by a ten
factor and this would be still relatively low.
 

Attachments

  • DSELF 2BTC THD20.gif
    DSELF 2BTC THD20.gif
    26.8 KB · Views: 1,101
  • DSELF 2BTC INV.gif
    DSELF 2BTC INV.gif
    8.1 KB · Views: 1,098
  • DSELF 2BTC NINV.gif
    DSELF 2BTC NINV.gif
    8.1 KB · Views: 1,087
www.hifisonix.com
Joined 2003
Paid Member
Good, we now have a more objective discussion going, if of course we can hold ourselves back from sleighting other people and their designs out of hand.

The input of TL's tone control stage does not need buffering. It's the opamp output that needs to be buffered. And, you can use jcx's neat to bias the OPA6x7 into class A.

Now, this arrangement would holds the prospect of of 0.005% distortion at 20KHz with full treble boost.

Wrt to the Baxandall tone control, it is well known that this must be fed from a low source impedance, and for this you can use an opamp or a suitable discrete circuit - there are many fine examples. Also, it is a very simple matter to arrange a bypass for a Baxandall such that is is 0 (zero) effect on the input signal.
 
Last edited:
Hi,

The input of TL's tone control stage does not need buffering. It's the opamp output that needs to be buffered. And, you can use jcx's neat to bias the OPA6x7 into class A.

Now, this arrangement would holds the prospect of of 0.005% distortion at 20KHz with full treble boost.

The arrangement I am suggesting needs no additional buffers, but optionally they MAY be used, should anyone percieve the need to do so..

There is no evidence that 0.005% Distortion under the conditions discussed is needed.

If we consider instead what is actually necessary we would be looking at 1-2V output with a few dB treble boost, which is the worst I would realistically expect the circuit to see and which will materially reduce observed THD.

However, if you MUST have such levels of distortion under such unreasonable conditions you may be required to add a buffer of some description. However this is not the same as "a buffer is required".

The suggested circuit will work with low noise and low distortion under normal conditions. And it allow all functionality that can be reasonably expected from a line stage for single-ended operation, using just ONE Op-Amp.

Ciao T
 
www.hifisonix.com
Joined 2003
Paid Member
Hi,



The arrangement I am suggesting needs no additional buffers, but optionally they MAY be used, should anyone percieve the need to do so..

There is no evidence that 0.005% Distortion under the conditions discussed is needed.

If we consider instead what is actually necessary we would be looking at 1-2V output with a few dB treble boost, which is the worst I would realistically expect the circuit to see and which will materially reduce observed THD.

However, if you MUST have such levels of distortion under such unreasonable conditions you may be required to add a buffer of some description. However this is not the same as "a buffer is required".

The suggested circuit will work with low noise and low distortion under normal conditions. And it allow all functionality that can be reasonably expected from a line stage for single-ended operation, using just ONE Op-Amp.

Ciao T

Three small signal transistors and 3 resistors are all it takes to get 10x the performance. Same applies BTW to the Baxandall.
 
www.hifisonix.com
Joined 2003
Paid Member
BTW I tried this topology in 1980 using very suboptimal LF356's to get around having to place a buffer in front of he Baxandall tone control. Of course, the problem for me in 1980 was I insisted on using a 100k pot. Now 10k is the norm and this opens up other options. I ended up sticking with the Baxandall.
 
Hi,

Three small signal transistors and 3 resistors are all it takes to get 10x the performance. Same applies BTW to the Baxandall.

I really personally prefer a TO-220 transistor plus LM317 (not for component count), but the principle result is essentially the same is the same.

However, we are still waiting for evidence that demonstrates that this level of performance is needed or that the improvement in measured performance translates into improved sound quality.

Ciao T
 
www.hifisonix.com
Joined 2003
Paid Member
You think DS's design is no good, or whatever term you used to dis it. The onus I think is on you to prove that your solution is superior, or that his is manifestly bad. I am simply pointing out that your generalizations are not based on fact.

The bottom line is a 30 cent solution is within a hairs breadth or perhaps better than a $10 solution. Not that the $10 op amp is bad, but it's real weak point in this specific application is not addressed in your proposal.
 
Hi,

You think DS's design is no good, or whatever term you used to dis it.

My point is that his design takes dozens of op-amp's to do what one Op-Amp can do (performance, function) and where the necessary circuitry has been been in the public domain for over three decades. As such the design to me has little to recommend it.

It is my personal experience (one that is shared by some) that the subjective sonic quality tends to be roughly inverse to circuit complexity, however this is another debate.

The bottom line is a 30 cent solution is within a hairs breadth or perhaps better than a $10 solution. Not that the $10 op amp is bad, but it's real weak point in this specific application is not addressed in your proposal.

I am not sure what weak point you refer to.

I have not seen any evidence presented by you that 0.005% THD at 20KHz and 7V output is reliably preferred or represents an actual, audible improvement over 0.05% THD at 20KHz and 7V when found in a line stage driving normal power amplifiers which in turn drive normal speakers.

Further, you have failed to provide actual data (not sims using questionable models) about the distortion of the circuit by D. Self. I did note this information in his article. I would suggest that the late great Bob Pease's adage applies here...

So whatever weak point you believe to have identified seems to be generated out of thin air in the whole cloth on several levels.

Ciao T