Page 2 of 5

Re: THAT1646 Current Booster and Transformer Driver

Posted: Sun Oct 04, 2009 3:47 pm
by mediatechnology
The nominal supply voltage is 27.5V, but it has to operate down to 17V.

To minimize power dissipation (so it will work at higher temperatures), I am trying to keep the supply at 7 volts.
OK. 7V rules out the 1646 since the minimum is 8V.

If the elevated temperature requirement didn't prevent it you could use the 1646 on 28V single-supply by using the differential output to directly drive the transformer primary and connecting a capacitively-bypassed rail splitter to pin 3. That biases the outputs to mid-supply producing an intentional 14V CM (not differential) offset floating both ends of the primary. See: viewtopic.php?f=6&t=261

You can also parallel multiple 1646s to split the dissipation since they have internal 25R ballasting resistors. The 1646 is also available on a SO-16W package with 80 deg C/W junction to ambient. It's the better package of the three for high temp. Still you may not be able to meet your elevated temp requirements though. It's not a world I often have to design for.

Re: THAT1646 Current Booster and Transformer Driver

Posted: Sun Oct 04, 2009 4:17 pm
by Jeffin90620
I'm not happy about high temperature design, either. I usually design automated test stations that run at room temperature.

It looks like the AD8139 will do the trick. If I keep the supply at 7V, it can run all the way up to 125C. Wish I had better SPICE models for real-world transformers, however; would like to see what the loading is likely to do before I fab the PCB.


Jeff

Re: THAT1646 Current Booster and Transformer Driver

Posted: Wed Sep 15, 2010 10:45 am
by emrr
I had need to put an active buffer/driver in front of a Pultec style mid-EQ filter, and I had these parts, so I put a 1246 in front of this 1646 configuration. It's working very well, and not breaking a sweat at any filter setting. I had two of this filter temporarily running with no buffer, single ended, just a 620 ohm input shunt as is normally found on the secondary of a 600:600 transformer. Ear comparisons with the passive input variation revealed no fingerprints from the 1246/1646, and filter action remains essentially the same. If there's any additional 1st filter gain range, it's very tiny. I don't see any need to use build-out R between 1646 and filter at all. I put them on those handy little Rat Shack 1.5"x1.5" IC perf boards, and floated them above the main board on a longer stand-off.

Re: THAT1646 Current Booster and Transformer Driver

Posted: Fri Sep 17, 2010 9:15 am
by emrr
Total available space was about 1" x 1" x 1.5" at most. There was no transformer that would fit the space and mounting, and do the job acceptably well, thus this solution.

I had a Pultec EQP-1A clone visit the studio, and was able to compare the Q of the high band LC filter with the MEQ type. I had been wondering about this for awhile. Will post some plots somewhere soon. Short story is the EQP type is much broader at the narrowest setting, and has much more available boost range. For example, if you set an EQP for the same max boost as the MEQ, for both narrowest and broadest settings, the shape of the EQP broad extreme is nearly the same as the narrow extreme. Key parameter being "for same gain setting". There is no verifiable calibration here, since varying the bandwidth control varies the filter gain separately from the boost control.

Re: THAT1646 Current Booster and Transformer Driver

Posted: Fri Sep 17, 2010 2:07 pm
by emrr
A 5534 and a 1646 would be another simple output combo here too, either driving an output transformer or not.

Re: THAT1646 Current Booster and Transformer Driver

Posted: Fri Sep 24, 2010 4:28 pm
by mediatechnology
emrr wrote:I don't see any need to use build-out R between 1646 and filter at all.
I think that the output impedance in the parallel SE drive configuration is 12.5 Ohms. So pretty darn low and essentially no buildout. Glad this worked out.

Re: THAT1646 Current Booster and Transformer Driver

Posted: Sat Nov 13, 2010 5:12 pm
by mediatechnology
Roger - You can indeed drop the 33R build-out to drive a transformer. It's there only for driving phones and hearing protection. Some phones in fact may sound better without it.

As a direct-coupled line driver using no transformer, the circuit is capacitive load stable as well without the 33R. The output buffer is running open-loop class-A and shouldn't oscillate. I tried it up to 1 uF before I quit testing.
Anyone?

I know Wayne has been really busy lately...
Yeah, my day gig remodeling the new home has kept my computer up there under painter's film and drywall dust for about the last week. That's the only part that drives me crazy. Sorry I missed the original question. Though computer used has been curtailed at least I still have running water...

Re: THAT1646 Current Booster and Transformer Driver

Posted: Sun Nov 14, 2010 11:09 am
by mediatechnology
Roger - Thanks.

I may need to clarify something for our readers. The circuit Roger is refering to is not actually posted in this thread but is the headphone amplifier with external driver transistors. For the sake of clarity it is here:

Image
THAT1646 headphone amplifier, transformer or line driver with buffer transistors.

The full thread here: https://www.ka-electronics.com/forum/ph ... p?f=6&t=23

The original drawings at the beginning of the thread work fine, but for even more output current, the circuit above "kicks ass" even more.

Re: THAT1646 Current Booster and Transformer Driver

Posted: Sun Nov 14, 2010 1:10 pm
by mediatechnology
Oops, wrong thread!
I think it fits here fine. One is for phones, one for transformers. Need to start one for driving cable. ;)

It's very adaptable circuit...

Re: THAT1646 Current Booster and Transformer Driver

Posted: Mon Apr 11, 2011 4:22 pm
by mediatechnology
Cool! Was it the buffered version or a bare 1646? Was there a build-out R?

If it's the headphone amp/buffered version you might be able to drop the 33R build-out. On the HP amps it's there for load short-circuit protection and with a tranny it could be lower.