1646 in a Distribution Amp

Construction information for DIY projects, including the MS Mid Side Matrix, Elliptic Equalizer, Mastering Console, Phono Transfer System, Insert Switcher and the Dual Class-A Amplifier. You can post your baby pictures here.
Post Reply
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: 1646 in a Distribution Amp

Post by mediatechnology »

Thanks for joining us.

I may have to give you some short answers since I'm under a bit of a deadline.

1) I would always feed a THAT1646 in that configuration directly from an op amp output. Any source impedance >100 ohm (and ideally <<100 ohm) will increase noise an decrease output balance.

2) I think keeping it unity gain bal in to bal out is probably the best option. If you were a radio station or TV plant you'd need trims. But you're "rebalancing" and for simplicity's sake just have the 1246/2143 feed the 1646s. But, realize that the input impedance of the 1646 is pretty low around 4K minimum. You really need a 5534-ish buffer between the 1646 and the 1246. So you do have an opportunity for an overall trim. You might want to make it inverting and then redefine the input polarity to keep it non-inverting.

3) The RCA ouputs I need to think about. If you take an unbalanced output off the 1646 you have to tie the opposing output to ground. One to get the +6 dB gain and also - this is very important - to keep the output noise down. An SSM, DRV or THAT line driver will all increase noise if an unbalanced load is not connected this way. The differential output noise on these parts is a lot lower than the common mode noise. This is due to CCOS SSM/DRV and OutSmarts. If the opposing output is left to float, the common mode noise will be converted to differential.

You might want to use a second 5534 or 1/2 5532 for the RCA output and give that one it's own level trim.

I guess they weren't so short after all. But it may be awhile before I can get back to you.
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: 1646 in a Distribution Amp

Post by mediatechnology »

Yeah what Roger said. He posted while I rambled.
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: 1646 in a Distribution Amp

Post by mediatechnology »

Yeah 4X 1646 would be a little more than 1K load on the 5534. If you need more than 4X 1646 outputs use a 5532 and have the 5532 drive two sets.
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: 1646 in a Distribution Amp

Post by mediatechnology »

Re-define it where? ....at the output of the 1646 ?
That might be the best place. You could do it at the input SSM2143. But your RCA outputs would be inverted. They see double-inversion as it is. Taking the + output of the 1646 and calling it - is about the easiest way to flip the polarity.

Overall your block diagram looks like a good starting point. (I think your SSM part number may be dyslexic.) You might want to put some low value 47-100R buildouts on the RCA outputs. The 1646s have them internally.

BTW thanks for posting it and joining us here. What tool did you use to draw it?
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: 1646 in a Distribution Amp

Post by mediatechnology »

Wow. M-S Paint. Never would have thought. I may try that as I haven't found anything as fast as a pencil yet for quick things.

I wouldn't do the RFI protection unless you're next to an AM antenna farm.

I would leave out output coupling caps but those 10 uF in the sense lines do keep the common mode DC voltage down. The whole thing can be DC-coupled if needed as the differential offset is typically 4 mV. Once the common mode offset hits a line receiver it gets canceled out.

The diodes aren't a bad idea but not absolutely necessary. On the Pico Comp we use little bridge rectifiers. If you're certain that phantom will never get patched into the output then it probably will. If you want to ditch the diodes and it does get hit then replace the 1646. The bridges sure make it easy. Diodes Inc. PN DB-104 comes to mind.

The main thing is to keep the inductance at pin 3 very low and take the bypass caps from Vcc and Vee directly to pin 3. I think I have a memory of an oscilloscope photo you sent.
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: 1646 in a Distribution Amp

Post by mediatechnology »

Hey you're welcome.

Good work. Those are nice gold pins BTW.
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: 1646 in a Distribution Amp

Post by mediatechnology »

Everything tests fine. I did notice the 1646 outputs had some DC offset, although within the datasheet specs, under +4 mV (with the 10uf sense caps connected). I was concerned at first, because it can be seen on the scope....a difference in level between the 1646 + output pins. I don't think this is a problem, right?
Good! That offset shouldn't hurt at all. If the OPA ahead of it is DC-coupled to the 1646, its Vos will be multiplied by two. e.g. if it's 3 mV it will be 3 mV at the 1646 + output and -3 mV at the - output for a total of 6. Then, add or subtract the 1646's own offset to that. Still it's very low and typical.
And, yes. Pin 3 (ground) of the 1646's needs to go DIRECTLY to the ground connection of that chip's 100nf power pin caps to prevent oscillation.... even an inch away may be too far.
They (THAT) should add this to their 1600 series datasheet.
It's on my list.
Post Reply