Transnova-Schade OS/Amp

Here is a nice and long-known circuit, (Jim Strickland's Transnova) with Schade feedback added (see the plot of the output characteristic in the attachment), transforming it into triode-like push pull output stage (or stand-alone amplifier with voltage and current gain).

What's nice about it:
- Great PSRR. Thanks to symmetrical topology the ripple waves from both rails cancel each other so when nicely balanced (no DC offset on the load) there is no noise on the output. Totally black background. No RC/LC filtering is needed (they actually make the situation worse - I put the 0R1 resistors in the rails just to check the current easier). Good and ample filtering capacity is all you need (with +/- 20V rails, at about 60,000 uF total per channel is when the low of diminishing returns starts to kick in).
- easy to work with. Lateral MOSFETs have Source on their body so mica or similar insulation is not needed, just connect heatsinks to GND. They also need no degenerative source resistors here, therefore enabling the square law mode i.e. extended class A current regime) and there are no thermal worries too - no temperature compensation is needed.
- High voltage swing, unclipped output is about 36Vpp on +/-20V PSU.
- Very fast and stable - see the 20 kHz square wave at the output with 5R load. Adding a 1uF in parallel to the load just makes the lines straighter with no ringing. I tested output level up to 95 kHz and it doesn't drop.
- Power supply rails are floating and each channel needs its own transformer with two identical secondaries so whether you want it or not you'll have dual mono amp. :)
- Clean, detailed and lively sound - no muddiness or harshness with complex orchestral music.

What's not so nice:
- Relatively high Zout - changing the load from 5 Ohms to 2.5 Ohms brings the Vout from 6.4 V to 4.5 V . It's not a problem if your speakers have high (8 or 16 Ohms) and steady (doesn't change a lot with frequency) impedance module. If not, you can encompass it into feedback loop with preferred gain stage or change your speakers... :)
- Relatively low Zin - as drawn it is about 1k (R1 || R4). R5 is there just for GND reference. Zin can be higher but at the cost of the bandwidth. In spite of that a lot of good preamps/gain stages will happily drive it: BA-3, LSK preamp, a lot of good OpAmps (AD829, AD797, to mention some).

How it works (attached schematic shows the test bed):
we see a complementary pair of Lateral MOSFETs, their Sources connected to GND and their Drains are AC-wise (through PSU caps) connected to each other, and load is connected between the Drains and the GND which makes it a push pull common source stage, giving voltage and current gain.
MOSFETs are biased through voltage dividers (R3+R2/R1 for Q1 and R6+R7/R4 for Q2). R2/R1 and R6/R4 relation is a local Schade feedback ratio and it sets the gain (R1 must equal R4 and R2 must equal R6 or you'll be introducing 2nd harmonic distortion). You can make this ratio higher or lower, depending on your preferences but you must have in mind that the bias current will need a new setting too.
R3 || C2 and R7 || C3 are DC shifters that enable us to separate biasing from gain settings (caps are short circuit for AC signal). As it is drawn, the bias current in real circuit is about 460 mA and DC offset at the load is 9 mV.
To adjust it at will change R3 and R7 into serially connected 15k resistor and 10k or 20k pot so you can finely adjust it to preferred bias current and 0V DC offset at the output (pretty much as described in F5 article - start with the max pot value i.e. minimum Vgs). There's not much point in biasing it to class AB although it will work...
Using a different rail voltage is OK, you just have to calculate your R3, R7 values anew (google: voltage divider calculator).
To increase the Zin, make R1 and R4 higher value but you'll have to change R2 and R6 accordingly (to keep the same gain ratio) and R3, R7 too (to keep the same bias current). Don't go too high with R1, R4 because it will lower the bandwidth (RC pole with MOSFETs' parasitic capacitance)

As you can see, the circuit is simple and rewarding. It opens a lot of space for experimenting and is highly accommodating to broad spectrum of applications.
Have fun with it...

P.S. If you still haven't grown out of "capacitor-in-the-signal-path" phase, this thread is not for you. :)
 

Attachments

  • tsdc.gif
    tsdc.gif
    24.3 KB · Views: 2,572
  • tssch.gif
    tssch.gif
    12.8 KB · Views: 2,653
  • tst.jpg
    tst.jpg
    111.9 KB · Views: 2,641
  • tsm1.jpg
    tsm1.jpg
    87.2 KB · Views: 2,294
  • Like
Reactions: 1 user
Nice and elegant design Juma - ) thank you !

Hi Kasey, have fun :)

Did square law improve output?

I'm not sure what do you mean by "improve output" but square law defines/describes the shape of the MOSFET's transfer characteristic. It enables the device to stay "longer" in the classA i.e. if we bias the device to 1A it won't switch off before it reaches 2 - 2.5 A.
I think that Mr. Pass wrote about it in one of his articles.

Nice to see you share a new design Juma! Will it replace your cubie in your case?

Hi Berny

Cubie has been replaced years ago - I've done a lot of other stuff since then. I started this thread just for real DIYers who like to experiment and make something nice of their own...
 
Last edited:
Hi Borko,
yes, I was exploring the Transnova common source topology and easily detectable benefits are: easier biasing - simpler circuitry, grounded Sources - no need to isolate Lateral MOSFETs from the heatsink.
Also, it most naturally fits the complementary folded cascode input stage (which I like the best). Since the output stage delivers very linear voltage gain the input stage is not overtaxed - it needs to swing only couple of volts for full power output (40 W / 4R in this case). Total result is exceptional.:wiz:
 
... a masterstroke indeed, Master-ji!!

Simple and elegant...

If you would be kind enough to "flesh out" #10 circuit with component values (for us foot soldiers!), we could warm up our soldering irons.

I would request more pointers for those of us wanting to understand the original topology and the Schade concept.

What will be the drive requirement for full output? An ordinary preamp would do, with say, less than 2V output?

Once again, big salutes to Master-ji...
 
Last edited:
Hi Prof,
I didn't publish the values because they depend on JFETs one has available (Idss varies in range from 2.6 to 20 mA - GR, BL and V). Although, with some effort, the circuit can be made universally acommodating. Anyway, there is enough data on this forum so that everyone can figure it out for himself (DIY is not just "solder it yourself", it's "learn it yourself" too).

Also, people tend to appreciate free design only if a big brand name or high price tag stands behind it. I got neither so ...

And most of all, I'm fed up with comments like:
"I have no idea what I'm doing but I simulated you design and it oscillates" ;
"I tested your design with wrong parts and connections and it doesn't work" ;
"I built your design poorly and it sounds like cr@p", etc...


So, if you find it interesting, put some effort in understanding it and come up with your own version.
 
  • Like
Reactions: 1 user
Thank you Sir for your kind response.
I do fully appreciate your stand and what all has prompted you to take up such a stand. Hobby fora, as far as I know, are not the friendliest of places. But my feeling is that for every naysayer and know-all, there are nine silent ones who appreciate the contribution of (gifted) members.

Personally, despite my limited expertise and experience, I dare say that your designs have an original "out of the box" quality that is plainly visible to even rookies. And they are simple, do not involve any exotic devices or complex topologies, and above all, they sound good...VERY good.

Sir, I believe I am speaking for many when I say THANK YOU FOR YOUR DESIGNS...WE ARE INDEBTED TO YOU.

I am in total agreement about the need for understanding and learning, especially in a hobby where we are trying to approach the rarefied regions in performance and the reproduction of realism. Being a solder-slinger has its limitations surely, but an ounce of guidance with some typical examples and circuit values, or maybe a hint at some comparable circuitry, will go a long way in bolstering the confidence of people like this poor self.

I shall set about wiring up #1 and make a start; then wait until the thread develops a bit and more knowledgeable people chip in, and perhaps you too condescend to "flesh out" #10, or a newer wrinkle on the theme, and then choose to go with that.

What will be the preamp drive requirement for #1 circuit? And kindly give some circuit comparisons and typical devices/values that you have tried for #10. Many of us would appreciate that certainly. (Kindly also ignore the know-alls.)

May the Force (of inspiration) be with you!!
 
Also, people tend to appreciate free design only if a big brand name or high price tag stands behind it. I got neither so ...

I don't think it's just me who would say that, around here, "Juma" is a fairly big name brand. ;)

Can I ask what voltage you used for the rails on the front end? Did you use +-15V transformer for the outputs like in the first schematic? Did you leave the bias at 460ma?