Jump to content

Discussion on the design of the T2 & other circuitry


kevin gilmore

Recommended Posts

So a couple of people have asked for a discussion of the design elements

of the T2, and any observations and improvements that are possible.

So as not to polute the T2 build thread, which is already seriously

poluted, the discussion of T2 design elements, including kerry's

dynamic amp should go here.

doug wadsworth (wadia) should be stopping by soon :D

Actually, a moderator change the thread title to

discussion of the T2, and other circuitry as we see fit

including but not limited to my other designs and

shunt regulators and stuff like that etc... :D

Link to comment
Share on other sites

Hello everyone,

I thought that I would get things going on this "design discussion" thread with an interesting little bit of the t2 circuitry.

Firstly, I must say that I'm very impressed with the circuitry of the T2. It's just so elegant in it's symmetry and it's use of valves or transistors, wherever they offer the best performance advantage. The dedication lettering that KG put on the PCB is most appropriate! However, I can't understand why Dr. Hayashi did the servo output in the way he did. He bounced the output current off the -12 V rail, using expensive (and hard to get) matched transistors. Why??? There seems to be enough headroon to use the + 12 V rail as a cascode point, making everything much simpler. Please see my attached drawing. So what am I missing?

I may be re-laying out the PCB of the t2, so anythng that we can thoroughly check out, that makes it less costly and easier to build, but that DOES NOT change the sonic characteristics of this classic product, will be considered.

I've got some (hopefully) interesting ideas on the Active Battery, so "don't touch your dial" :) Stay tuned to this thread.

Bye the way, just to clarify what KG mentioned a message or so back, I did do design work for Wadia (analog "swift current" section of the 861 CD player), but I was a consultant and not an employee of the company.

Linear

post-2933-0-61264700-1301018453_thumb.jp

Link to comment
Share on other sites

Well, first these transistors were not hard to get and certainly not expensive at the time. I do like the change though.

Good point, Luvdunhill. But that was then and this is now! Now, Bdent want $7.66 for the 2sc3381 (anyone found a better price?) and you need FOUR of them to make the simple cascode current mirrors used on the servos of the t2! You can buy all the HV NPNs you need for the t2 at that price!

But, in keeping with the "design" theme of this thread, I submit, with the utmost respect and admiration for Dr. Hayashi, that it would be completely out of character for him to use up 3381s just because they were cheap and available. The design of the t2 is concise and elegant. I believe that he had a good technical reason for using that topology. Comments?

Link to comment
Share on other sites

Good point, Luvdunhill. But that was then and this is now! Now, Bdent want $7.66 for the 2sc3381 (anyone found a better price?) and you need FOUR of them to make the simple cascode current mirrors used on the servos of the t2! You can buy all the HV NPNs you need for the t2 at that price!

But, in keeping with the "design" theme of this thread, I submit, with the utmost respect and admiration for Dr. Hayashi, that it would be completely out of character for him to use up 3381s just because they were cheap and available. The design of the t2 is concise and elegant. I believe that he had a good technical reason for using that topology. Comments?

Yeah. 2SC3381GR

Caveat: Have not tested mine yet.

Link to comment
Share on other sites

Great idea for a thread since I need the T2 thread to rant about my PSU... :D

The dedication to Dr. Hayashi was the least we could do to show our admiration for all the work he's done. Same goes for his father as without his groundbreaking work we would still be using inferior dynamics... :)

Link to comment
Share on other sites

The simulation shows absolutely no difference between the 4 transistor

current mirrors, and the single transistor circuit. Does need to be

a 100 volt transistor however. Due to tube shorts, probably should

be rated at 300 volts. Whatever the npn equivalent of the mpsa92

should work great.

Thanks for running the sim, KG. It was on my "to do" list, but right now I'm testing heat sinks and doing sims on the active battery. (I think that I may have really got somewhere with the battery .....eliminated the 2sc3381 and more.....stay tuned.)

The 3381's are only 80V, which is probably why there is a 10K in series with the mirror collector (to drop 5 more volts and protection). We have almost 24 volts LESS across the transistors, but there is a downside to this. Is everyone SURE we still have enough HEADROOM???? Of course, KG is right - Put a hv NPN in for maximum safety!

Regarding thermal stability, Luvdunhill, I think that the cascode (off the +12 V rail) is better. The output of the servo opamps is a voltage, converted to a current with the 22K resistors (R84 & R85). There are 2 Vbe in series with these resistors in the original version, and only 1 with the cascode. (We can match the original by adding a diode in series with the cascode, if this additional Tc has some magic effect. :) )

And now the BIG QUESTION: Who will volunteer to try this out in his T2? :o

Linear

Link to comment
Share on other sites

Actually the 2sc2240 would work just fine.

But i was simulating absolute worse case.

Lets say the current source on the top of the top input

tube shorts, then the top tube shorts... Suddenly

250 volts on the cathode. Of course this has

not happened yet, and is unlikely to ever happen.

But i do wonder what would happen if just the tube

shorted. This scenario would blow up the 2sc3381's

anyway.

And this is why i figured it would be easier to use

the 300 volt versions. Which are still dirt cheap.

Edited by kevin gilmore
Link to comment
Share on other sites

does the offset op amp circuit need a common-mode servo/control?

at a 1st glance it looks to me like it would be perfectly happy with one op amp integrating to a rail on one side with the other then working "single ended" to servo the offset

Link to comment
Share on other sites

The common mode servo is Q34,Q35 and Q36.

You could replace those, but you would need

another opamp to do it.

Even though you could get rid of a few parts

you still would need an inversion from the

one opamp to drive both current mirrors.

So you need 2 opamps for that.

Edited by kevin gilmore
Link to comment
Share on other sites

Everything in the servo circuit is relatively inexpensive and available, except for those 2sc3381's! That's why I looked long and hard to see if they were really necessary. Apart from the cost, it's still probably a good idea for them to go - I think they are going to get the "last time buy" notice very soon, if they haven't already. (Maybe bdent is selling NOS).

Anyone out there brave enough to try the cascode mod in their T2? :o But you do so at your own risk! B)

Link to comment
Share on other sites

I'm still guessing that U7,8 op amps outputs common mode V is unconstrained and its possible the circuit "works" with one of them saturated at a rail

I would just use one differential integrator and turn the 2nd op amp into a gnd referenced inverter of the 1st's output

then a resistor from each op amp out to the 2 cascode Q would give offset currents that are symmetric about a common bias value

Edited by jcx
Link to comment
Share on other sites

Actually if the batteries are adjusted well, the opamp output voltages are no more than +/- 3 volts.

You could certainly turn the other opamp into an inverter and drive the current mirrors that way.

A savings of a few parts.

edit: if one of the opamps was at the rail, the other opamp would have to be at the

other rail.

Here is the first cut of the hybrid shunt regulator version of the kgsshv power supply.

http://gilmore.chem.northwestern.edu/shuntregver2.pdf

Unlike some other shunt regulators however, the current sources

shut down if the shunt element pulls too much current, so its

safe to power up with no load.

edit: silly bench mistake fixed. Actual noise under load about 3.5 microvolts peak to peak.

Edited by kevin gilmore
Link to comment
Share on other sites

sure is much simpler. I wonder how stable it is.

Will evaluate :D

my model for the tl431 is probably lousy, but

load regulation is in the many percent, and the

noise is 7 or 8 millivolts. You are relying on

the pass fet to stay stable over temperature, and

that certainly is not the case.

Will play more :D

can host ltspice stuff.

Edited by kevin gilmore
Link to comment
Share on other sites

I wonder how stable it is......

load regulation is in the many percent.....

noise is 7 or 8 millivolts......

pass fet (not) ...... stable over temperature....

It mignt need a cap (eg 0.01 uF) across the cathode and ref. (See Texas Instrument's data sheet, figure19.) Or perhaps the Linear Technology version, LT1431, which has a compensation pin (see pg 8 of their DS).

For most amps, we probably don't need the world's best PS, just something that's reasonable good, and simple and cheap to build. Hell, with TL431's, it will be way better than the original T2 PS! (My apologies, Dr Hayashi. Your father probably cut the budget for the T2 PS after you used up most of the project time getting the amp section virtually perfect!).

Link to comment
Share on other sites

OK, this is the design thread, right? Anything goes! :lol:

So let's go hog-wild with 431's, and replace the series pass devices with something in an isolated case.

Here's what I came up with. Note the use of the TLV431 (it's the TL431's baby brother). It can regulate at 100 uA (versus 1 mA), and that saves us power in the bias resistor. The NPN version might be OK, but has the Vbe Tc. Why not use more 431's ! :lol:

post-2933-0-01318500-1301353633_thumb.jp

Link to comment
Share on other sites

Here it is:

Not the world's best PS.

But simple, easy to build (all isolated packages), and inexpensive.

And it has KG's novel hybrid feature.

I threw this togther rather quickly, in the last few hours, so there may be a fatal flaw! Feedback appreciated.

Linear

post-2933-0-05233000-1301358363_thumb.jp

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...

Important Information

By using this site, you agree to our Terms of Use.