THAT1646 SSM2142 Balanced Ouput Driver Comparison

Thanks; looks interesting, and not too expensive (compared to getting an MOQ worth of '1606s and toasting a few while soldering). The only downside I can see is that it lacks the recommended heat sink copper areas.

JDB.

It was only intended as a "prototyping board" for Graham to decide if the 1606 would be suitable for a particular application which was made because of the inpracticality of trying to wire up these devices..... and we never put ground planes onto prototype boards (because it has no soldermask)

I'll put a detatchable ground plane on the next revision.

Colin.
www.audiomaintenance.com

I just used a bunch of 1646 in a balanced buss mix buss fab, they worked great but they are NOT pin to pin drop in replacements for a SSM 2142! Standard practice on a 2142 is to tie the sense and force lines together then use a coupling cap to block DC, the THAT chips do not like this at all. They do work fine with the 10 uf cap between sense and force, BTW I used Nichon Bi Polar / lytics for this

most of the gear I have worked on that actually use 2142's (Amek has used them extensivly, just to name one..) tie the force and sense lines together and then use a polarized electrolytic, often 220uf, on the outputs. When I tried the 1642 in this way on the balanced mix bus, there was a drastically high noise floor. I would expect problems if they were used as a drop in replacement for the 2142's. They are also very sensitive to power supply high frequency noise, and must have 100n decoupling caps very close (not so with the 2142's)
That being said, with these devices implemented as per the manufactures spec, they work fantastic

[quote author="nielsk"]most of the gear I have worked on that actually use 2142's (Amek has used them extensivly, just to name one..) tie the force and sense lines together and then use a polarized electrolytic, often 220uf, on the outputs. When I tried the 1642 in this way on the balanced mix bus, there was a drastically high noise floor. I would expect problems if they were used as a drop in replacement for the 2142's. They are also very sensitive to power supply high frequency noise, and must have 100n decoupling caps very close (not so with the 2142's)
That being said, with these devices implemented as per the manufactures spec, they work fantastic[/quote]

nielsk,

I am curious about the problems that you encountered. I would not expect any problem with shorting the outputs to their respective sense terminals (in fact, the parts are tested this way), or with output AC coupling capacitors. Your comment about increased noise makes me wonder about the source impedance driving the part. All of these designs have inherently high common-mode noise gain, and an imbalance in the feedback network will convert common mode noise to differential noise. The 1646 is more sensitive to this than the 2142 or DRV134, in part due to the lower input resistance (which yields lower noise when they are driven with a 0 ohm source). We have found that some customers, despite the cautions in the datasheets of all three parts to drive them from opamp outputs, drive the part via a higher source resistance. This may make what was an acceptable cost/performance tradeoff (saving an opamp) with the SSM2142 or DRV-134 unacceptable with the 1646.

As to the 100 nF decouping capacitors, yes they must be close to the IC, as is the case with most high-speed devices. You may have seen stability issues if they were too far away. I would expect HF power supply rejection to be at least as good as the competitive parts.

If you could let me know the details of the circuit that gave you problems, we can investigate a bit further. Feel free to PM me if you like.

Regards,
Gary Hebert

It was fed a signal buffered by a 5534 at unity gain, post fader. This signal was fed to the L/L+R/R switch, (grounding the un used side) via a 220uf cap & 10K to ground/10 ohm resistor in series, then directly to the 1646 input. The outputs fed a 15K resistor, mix bus going to 1 & 8 of a SSM 2017, with 2&3 grounded. Then the signal went to a 1646 via a 220 uf cap to create the mix insert send.
I am currently working on a Pan circuit (probably the two 10K lin pots tied to ground in the middle approach) and trying to get a sample of a 1510 / 1512 to try in place of the 2017, as well as a servo for the mix amp output. Any ideas on improving this and using THAT chips would be greatly appreciated!

[quote author="nielsk"]It was fed a signal buffered by a 5534 at unity gain, post fader. This signal was fed to the L/L+R/R switch, (grounding the un used side) via a 220uf cap & 10K to ground/10 ohm resistor in series, then directly to the 1646 input. The outputs fed a 15K resistor, mix bus going to 1 & 8 of a SSM 2017, with 2&3 grounded. Then the signal went to a 1646 via a 220 uf cap to create the mix insert send.
I am currently working on a Pan circuit (probably the two 10K lin pots tied to ground in the middle approach) and trying to get a sample of a 1510 / 1512 to try in place of the 2017, as well as a servo for the mix amp output. Any ideas on improving this and using THAT chips would be greatly appreciated![/quote]

Why did you go first to balanced with 1646, and then to 2017? Could you just feed unbalanced to 2017 mixbus?

You mean servo on mix amp like this?




...

[quote author="nielsk"]It was fed a signal buffered by a 5534 at unity gain, post fader. This signal was fed to the L/L+R/R switch, (grounding the un used side) via a 220uf cap & 10K to ground/10 ohm resistor in series, then directly to the 1646 input. The outputs fed a 15K resistor, mix bus going to 1 & 8 of a SSM 2017, with 2&3 grounded. Then the signal went to a 1646 via a 220 uf cap to create the mix insert send.
I am currently working on a Pan circuit (probably the two 10K lin pots tied to ground in the middle approach) and trying to get a sample of a 1510 / 1512 to try in place of the 2017, as well as a servo for the mix amp output. Any ideas on improving this and using THAT chips would be greatly appreciated![/quote]

Hi Nielsk,

I don't think that it was the 10 ohms. This barely budges the output noise. 100 ohms would increase the noise floor by about 2.4 dB. The 220 uF capacitor does not significantly contribute to the source impedance,so I'm not sure what's going on here. What noise level were you measuring at the 1646 output? BTW - I assume that the switch in series with the input was an electromechanical type rather than a solid-state switch.

Regards,
Gary

[quote author="mediatechnology"]
Moral of the story is drive all three of them, 1646, 2142 and 134 from a (1) low resistance source,[/quote]

what about the 1606? I was hoping to be able to use it as an input buffer. I have samples but the package and lack of time to deal with it have kept me from experimenting. maybe this weekend.

mike

thanks Roger, but this is not what I am trying to do. Im trying to eliminate a transformer, I need to make a circuit with near unity gain, balanced input and balanced output. the output balance is the most important spec, then input CMR. a transformer does quite nicely, but my whole motivation is to revive a design I had to put on the shelf because it was getting too expensive to build. I did make some progress on this but I still havent managed to get those 1606's wired up to verify.