THAT1646 Current Booster and Transformer Driver

Where we discuss new analog design ideas for Pro Audio and modern spins on vintage ones.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: THAT1646 Current Booster and Transformer Driver

Post by JR. »

Yup, I was thinking of something other than the old drive the speaker as a current source, but that works too.

I was more thinking in terms of using DSP for smart loudness compensation, and drive protection in large sound reinforcement applications. Actually measuring the VC temp, or one of several indirect methods are available when have the dots to connect.

JR
Cancel the "cancel culture", do not support mob hatred.
ricardo
Posts: 262
Joined: Sun Apr 10, 2011 7:24 am

Re: THAT1646 Current Booster and Transformer Driver

Post by ricardo »

JR. wrote:... something other than the old drive the speaker as a current source, but that works too.
That was investigated by Mills, B&W and does reduce distortion but so does ACE-bass

Synthesis of Loudspeaker Mechanical Parameters by Electrical Means - Karl Erik Stahl
A method for extending bass response and lowering distortion in louspeaker systems, which differs from both equalized and servo or feedback systems, is described. It can be seen as an extension of the technique to increase loudspeaker damping by giving the driving amplifier a negative output resistance. However, this method also controls the moving mass and compliance. Very good results, such as flat frequency response down to 20 Hz and excellent distortion data, have been achieved with moderately sized bass-reflex enclosures.
AES Convention:61 (November 1978) Paper Number:1381

Current drive is MUCH more complex to get right.
.... actually measuring the VC temp ...
This is what David's bridge did. I think he never had a preprint (Inst. of Acoustics, Windermere 198?).
ACE bass is the best method I've tried and I had some success with incorporating David's VC temp compensation in it.
Low distortion transformers seem counter productive for all the folks who use them solely to add coloration.
Golden Pinnae Brigade aside, I think there are still good solid reasons for transformer in/out. With all respect to the excellent THAT stuff ..
- better, easier real-life RFI bombproofing.
- easy P48V
The best I've managed with transformer mike i/p is about 1.6dB NF but this is only important with a low output ribbon like Coles 4038. For that, a proper Cohen (or even THAT 1510) without the P48V bits but full RFI, would be slightly better.

For output, you rarely NEED more than 22dBu (comfortable NE5532 on +/-15V) I don't think the fancy balanced outs are worth the PCB space. I'd rather use "Balanced Impedance". When you DO need more than 22dBu into a low load, it is hard to beat a transformer as in the earlier Lundahl note.

If only Swedish virgins and Unobtainium cores weren't so rare, large & expensive.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: THAT1646 Current Booster and Transformer Driver

Post by JR. »

ricardo wrote:
.... actually measuring the VC temp ...
This is what David's bridge did. I think he never had a preprint (Inst. of Acoustics, Windermere 198?).
ACE bass is the best method I've tried and I had some success with incorporating David's VC temp compensation in it.
Yup... in the fixed install business the expensive amplifiers for computer controlled systems often include output current sensing. They use them more often to detect open or shorted drivers, say in an amusement park, but comparing cold current draw, vs hot current draw for a given output voltage also gives you a handle on temperature rise from VC resistance change.

For a while I considered selling a stand alone protection box that went between amps and speakers, with my fancy peak/VU meters for blinky lights appeal, but decided it was too much trouble to hand hold consumers to set it up properly for sundry speakers, and then they would hold me responsible when they still blow up any speakers.

I was also looking into a related mechanism to detect power transformer winding overheating. In that case the resistance change from winding temperature rise, could be parsed out from the voltage drop at the peak of the unregulated supply ripple. The Vp-p of the ripple gives you average current, and unreg voltage at say 45' can impute what unloaded voltage "should be". I never reduced this to practice, because it was generally cheaper to use a bigger transformer, or put a thermal one shot inside, but again inside a powered speaker, if you have computing power to burn, you could do the math and protect VC and transformers, while these days it's hard to imagine not using a switcher inside powered speakers that already have a weight problem.
The best I've managed with transformer mike i/p is about 1.6dB NF but this is only important with a low output ribbon like Coles 4038. For that, a proper Cohen (or even THAT 1510) without the P48V bits but full RFI, would be slightly better.
I am a little uncomfortable crediting Cohen with that preamp topology but lots of the early audio work was not published or widely disseminated. Discussion in another thread here.

Input transformers are not needed for blocking phantom voltage if you float the whole preamp up to nominal phantom voltage. See Waynes design in another thread here.

I will concede transformer's general superiority at blocking RF, but there are alternate inductive input filter approaches that just block the RF, without forcing the audio frequency content into the magnetic domain and back again.

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: THAT1646 Current Booster and Transformer Driver

Post by mediatechnology »

There's alot to be said for galvanic isolation.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: THAT1646 Current Booster and Transformer Driver

Post by JR. »

I used lots of transformers in mic preamps for the fixed install market, but they were some pretty low cost stuff... I had to do some leg work at first sourcing transformers from China that were even usable, since the sources I had didn't even know what mu metal was, :oops: They had trouble sourcing that locally, but we finally got some parts that worked.

That install market even insisted on full galvanic isolation for the 70-100V output power transformers. We could have used much cheaper auto-formers for the output voltage step-up and saved a bunch of iron and copper, but the customer is always right. :roll: My suspicion is they liked the true transformer outputs, because they could inadvertently short circuit one side or the other of a speaker line to ground and the system would still work, saving them on skilled installation labor, or an expensive service call.

Galvanic isolation for mic input transformers seems over rated (IMO), since phantom powered mics won't work without a ground return connection. For live sound reinforcement, there is perhaps some merit for line output transformers to ignore tangled up mains power distros, and ground potentials between FOH and stage power, but unsafe mains power should not be tolerated anywhere, IMO.

JR
Cancel the "cancel culture", do not support mob hatred.
ricardo
Posts: 262
Joined: Sun Apr 10, 2011 7:24 am

Re: THAT1646 Current Booster and Transformer Driver

Post by ricardo »

JR. wrote: I am a little uncomfortable crediting Cohen with that preamp topology but lots of the early audio work was not published or widely disseminated.
By "proper Cohen", I meant a practical mike pre using a similar topology, full RFI & P48 protection, with well under 1nV/rtHz noise. Scott Wurcer pointed to Demrow which I think is the earliest mention but not in a low noise context.
There are many Cohen copies, most not even coming near to THAT 1510 etc. done properly. I'm sure you know that approaching a chip or circuit's true potential is all in the details.
Millenia Media use the "Cohen" as probably Earthworks. The "proper Cohen" is the lowest noise mike pre. in the known universe and the Earthworks the lowest noise you can buy.
I came to this forum from micbuilders discussing this and posted both Wayne's floating PSU solution and his "new" offset method for THAT 1510 etc. The latter is probably what I'd use for a commercial pre. But if cost & space was no object, I'd still use a Lundahl mike transformer & sacrifice 1dB of NF. I believe in counting bits for any (full RFI & P48 protection) solution. A good trannie reduces loadsa stuff. Even allows clickless gain switching w/o Unobtainium caps. The floating PSU is potentially better ... but do I really want special PSUs for each channel?

What I want to know is if a "proper Cohen" built with his Philips (Oz) hybrid chips is clickless even w/o the Unobtainium cap. Does the Millenia Media have a cap? No point looking in your Earthworks cos the good stuff is encapsulated.

Maybe this question belongs in another thread.
.. but comparing cold current draw, vs hot current draw for a given output voltage also gives you a handle on temperature rise from VC resistance change.
David used a DC bridge with a small DC offset on the power amp.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: THAT1646 Current Booster and Transformer Driver

Post by JR. »

The last low noise preamps I designed were all before IC preamps were truly competitive, and while the extremely low Rbb (moving coil head amp) devices were still being sold. .35-.4 nV rt/Hz was adequately low noise, but DC offsets with discrete devices are another story.

The THAT chips are quite good with 250uV DC offset voltage, but the offset current into a practical DC termination impedance will dominate the "click/scratch" factor.

Their app sheet shows a cap coupled gain pot. I wonder if they could integrate in a DC servo?

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: THAT1646 Current Booster and Transformer Driver

Post by mediatechnology »

Roger - I think the buffered class-A might have some merit in non-transformer coupled applications where long cable lengths and lots of capacitance are anticipated. I agree though that it's hard to improve on the 1646 connected directly to a transformer. To get the output impedance really low for the buffered circuit though it needs some short-circuit protection. I depend on the 27-33R to do that for the headphone amp.

With 2X 1646 paralleled as you did the build-out output resistance is 12.5R. That's pretty low. I've been curious to see just how many 1646s could be paralleled. I stopped at 4X outputs (2 1646s).
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: THAT1646 Current Booster and Transformer Driver

Post by mediatechnology »

RF - Thanks for the update. I'm moving right now so my response is a bit delayed.

While daydreaming about electronics the other day I had thought about this very circuit where multiple devices were paralleled and used to drive headphones directly. (I think I was daydreaming while mowing :D )

You think you might be able to clip a pair of headphones onto the output of your transformer driver using 4X THAT1646 and see how it sounds? I think the Sony MDRs around 32 Ohms and older 500 Ohmish AKG 240s might be a good test.

With 4X 1646 the output buildout R is around 6.25 Ohms.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: THAT1646 Current Booster and Transformer Driver

Post by mediatechnology »

Nice work Roger. Thanks for posting this.

Available in Red: http://www.gearslutz.com/board/new-prod ... rsion.html
Post Reply