Who designed the Radford ZD22 preamp?

[ginetto61]

So, what do you need then? (In terms of gain, noise, source impedance, load impedance and levels.) Eventually you'll start running into problems at low gain with this circuit, with noise possibly not being as low as you may like. At some point you start having to add an output pull-down resistor or current source.

Hi ! the idea is to use the design for a line stage ... so gain voltage of 3-4 times max and load 10K and up ...

As i am focusing only the output stage of the ZD22 i have to reduce the value of the feedback resistor to get the gain of 3-4 (I am cutting out the input amp stage and phono stage ... i do not listen to LPs anymore ... only their digital copies at max)

A friend of mine has already done that with no issue One thing he told me is that changing the feedback resistor for an Holco opened up the sound a lot. Even the small details count.

Good idea. As long as they aren't too busy going extinct again, that is.

My parts selection will be limited by the bjt models available because i really want to simulate first the circuit, given that is very simple it could be a nice learning step. And see the impact of gain, voltage supply, different bjts, and so on ... Once i get very good distortion and noise graph i will start building a prototype ... that is the plan

A key to the good performance of these rather simple circuits is their high supply voltage. It saves you the odd current source (while still allowing high loop gain) and keeps Cob low. The downside is that some desirable parts top out at 45-50 V C-E and may require cascoding.

Thanks a lot for the very valuable advice. Changing topology is absolutely out of question. I am sure this already good design can provide very very good sound when optimized. So better spend time optimizing a very simply topology than going the complexity route. The scope is to get an excellent gain stage usable as line amp/buffer with gain with just two bjts.

For some further refinements you can check out my 3 transistor design. It uses some more refined input biasing that gives better PSRR (at the cost of adding one resistor and capacitor per input), and the more sensitive and not very current-hungry input amplification circuitry got its own little RC power filter that doesn't have to feed the output buffer. One thing I never really got worked out entirely is the matter of turn-on thump. The output necessarily has to rise to about half supply or around 24 V in this case. It works out OK if I use a 2µ2 output capacitor and 10k of pre-loading, but I'm not entirely happy with that. Perhaps going 2nd order would help, otherwise a delay / muting circuit may be required. This really is a drawback of just about any single-supply circuit if you also want some serious low-frequency extension

Very interesting and i am sure that your project is much more refined. However as i said i have an open issue with this circuit. I have read so many people praising the sound of the original that can be even not optimized for just line preamp duties. I have to find a source for bjt models and start simulating. I will ask about that Thanks a lot again.

 

[sgrossklass]

Here's a little sim for a Gv = 4 circuit. You need Bob Cordell's models for the BC550, the BC556 model ships with LTspice I think. It's also right at the BC550's Vce limit with a 55 V supply - the output side is less critical.

It appears that the output side will benefit greatly from transistors with high Early voltage Va and low Cob. The likes of 2N5401, 2SA1514K or MPS8599 should work well. Note: Beware of transistor models with "dummy" VAF, such as exactly 100 V. Also note that varying DC bias you may find a "sweet spot" in distortion that may be largely cancellation of 2nd harmonic and not actually as significant in real life.

You can take some voltage off of the input transistor by adding two extra components, but beware that your optimum emitter resistor value is likely to shift. That will also depend on the frequency you're looking at, as towards 20 kHz you increasingly run into slew rate issues and slew rate is also influenced by this resistor.

 

[ejp]

Changing topology is absolutely out of question. The scope is to get an excellent gain stage usable as line amp/buffer with gain with just two bjts..

I don't understand. The original topology uses seven BJTs. If you're only going to use two, there isn't anything left of the original topology except a CFP with gain.

 

[ginetto61]

Here's a little sim for a Gv = 4 circuit. You need Bob Cordell's models for the BC550, the BC556 model ships with LTspice I think. It's also right at the BC550's Vce limit with a 55 V supply - the output side is less critical.

Hi ! thank you very much indeed for your precious support.
My main object of interest is the output stage only.
As i said before i am very uneducated in electronics ... but i would like really to select the best parts (i.e. active and passive) for the job.
My point is:
1) i really do not know how to do it
2) i am afraid that specific models for those best parts will not be available ?
I am still stuck on point 1.

It appears that the output side will benefit greatly from transistors with high Early voltage Va and low Cob. The likes of 2N5401, 2SA1514K or MPS8599 should work well. Note: Beware of transistor models with "dummy" VAF, such as exactly 100 V. Also note that varying DC bias you may find a "sweet spot" in distortion that may be largely cancellation of 2nd harmonic and not actually as significant in real life.
You can take some voltage off of the input transistor by adding two extra components, but beware that your optimum emitter resistor value is likely to shift. That will also depend on the frequency you're looking at, as towards 20 kHz you increasingly run into slew rate issues and slew rate is also influenced by this resistor

i wonder if a simulation can reveal all this ... if not it will be a big problem.
Again the circuit employs only one NPN and one PNP.
When i see this i feel lost ... the choice is almost endless ...
Transistors | Products | Toshiba Electronic Devices & Storage Corporation | Americas – United States

maybe these could work ?

Transistors | Products | Toshiba Electronic Devices & Storage Corporation | Americas – United States

for the PNP part that i remember gets hotter i would prefer a part easy to heatsink properly. And more powerful ... like this one

HTTP 301 This page has been moved

I really believe in the CFP topology because it was very popular in the 70s in high end preamp like also a Kenwood 700C for the line buffer block
With the best parts now available, with the right simulation i think a new circuit could outperform even the original

But between to say and to actually do there is a sea ... we use to say here
i am not that good at sailing ...
Thanks a lot again

 

[ginetto61]

I don't understand. The original topology uses seven BJTs.
If you're only going to use two, there isn't anything left of the original topology except a CFP with gain.

Hi ! yes indeed ... that is just what is needed for a line preamp.
If there is one thing that i do not need at all are tone controls ... and i understand the tone controls are the reason for this complicated schematic ?
by the way .... it seems that CFP is not that popular these days

 

[ejp]

Well there are always better topologies, even within the CFP world, for example a CFP with CCS, but I'm struggling to see what all this really has to do with the Radford topology, which is indeed 2/3 about driving a tone control stage.

 

[ginetto61]

Well there are always better topologies, even within the CFP world, for example a CFP with CCS,

Hi ! thanks very interesting. May i ask in which sense they are better ? lower in distortion maybe ?

but I'm struggling to see what all this really has to do with the Radford topology, which is indeed 2/3 about driving a tone control stage.

yes ... i am just interested in the othe 1/3 that drives the power amp stage. That is my object of interest. A very simple but good sounding line stage.
I have seen line stage with tenths of active parts and even more passive parts.
Is all that stuff really needed ? for what ? is it not a case of overengineering ?
why not optimizing simple topologies to their limits instead of take the complexity route ? a KISS principle application ...
I am just pruning out what is unnecessary for me ... like tone controls.