THAT 1570 1510 and 1512 Input/Output Circuits

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: THAT 1570 1510 and 1512 Input/Output Circuits

Post by JR. »

Sorry to jump into the tail end of a conversation.

Image

+1 to the suggestion to add LPF caps across R2 and R4. These caps need to be same value and similar precision to Rs for HF CMR, while 1% caps are surely rare.

R7 is only useful for compensating nominal DC offsets with old school bipolar opamps that had significant input bias current (dating that schematic). Modern bipolar opamps with bias current cancellation, or FET inputs with << bias current, receive no benefit so you can lose the extra noise source, however modest.

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: THAT 1570 1510 and 1512 Input/Output Circuits

Post by mediatechnology »

Thanks John.

A couple of other points:

There is no 49R9 in the -Vout leg on the ADI drawing.

The junction of R7 and the non-inverting input is a convenient point to inject a Vcm. This is handy when the circuit is made attenuating and (Cload tolerant) for A/D driving.

BTW the noise contribution of U2 is almost "free" being that it essentially appears in common mode at the outputs. If the following stage is differnetial and has good CMR U2s noise contribution is very small.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: THAT 1570 1510 and 1512 Input/Output Circuits

Post by JR. »

I pretty much ignored the 49 ohm R since I never found much use for a balanced input to differential output circuit. I typically used that topology to single end a balanced feed coming into a console or rack gear, and since it would be lightly loaded by the following stage there is little need for build out R. The attraction of this circuit is executing a balanced input using only two opamps, so if we plan to budget using three opamps, we could just use a true instrumentation amp topology in the first place and gain higher input impedance, lower NF network resistance, etc. The true inst amp topology being more symmetrical and naturally balanced, would behave better at frequency extremes too. Note: There is an obscure and somewhat esoteric benefit from using opamps inverting, as input stage LTP CM issues are not exercised as strenuously. So this topology may deliver some unexpected benefit with lesser opamps, over even pure instrumentation topology.

Visualizing this circular feedback path is interesting. If you look at the -Vout node as an input to U1, assuming Vin + and - inputs are terminated to a low impedance, it will look like another + input to a unity gain non-inverting stage. So opamp U1 will try to make it's output node follow the voltage at -Vout (it's effective + input). This makes R5 and R6 appear in parallel as feedback to U2's - input helping it follow the U2 + input (noise). So while not very intuitive that noise will be coherent and common mode at the input and output of the U2 inverting stage, and cancel in a following differential. Of course this analysis ignores lag, feedback network poles, etc. making caps across R2 and R4 identical will maintain the unity gain relationship wrt -Vout. I vaguely recall some discussion about putting a cap across R5 = cap across R6. Of course this is not generally accepted practice, so I would keep these caps tiny, and make LPF poles across R2 and R4 dominant. Having some C across U2 helps U2 present a low impedance at HF to the outside world via the input and U1 feedback network. I always like to visualize some radio signal in input ports. Of course a another stage of passive LPF could be added in series with inputs terminated to actual ground, for robust RF rejection.

Since that following differential would add noise of it's own there are diminishing returns in adding a stage to cancel such a modest noise source.

have we beat this one to death yet?

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: THAT 1570 1510 and 1512 Input/Output Circuits

Post by mediatechnology »

have we beat this one to death yet?
No. :D

I suppose I should show the A/D driver using it.
hazmo
Posts: 19
Joined: Wed Dec 23, 2009 7:31 am
Location: Germany

Re: THAT 1570 1510 and 1512 Input/Output Circuits

Post by hazmo »

Well, thanks for beating it, I never can't get enough of it :lol:.
They don't tell you stuff like that in lectures...


In the meantime I made a preliminary drawing for the preamp controlller. Since the plan is to have a remote preamp,
there will be one board that directly talks to the THAT5171s, and one board providing the hardware interface for the
user, both connected by I²C. Something like 50 meters or more will be possible with the bus drivers I want to use,
actual maximum distance depends on the environment. One set of boards will be able to control at least four preamps I
think, depending on the feature set of the user interface I haven't fully decided on yet, it might be up to eight. The
prototype will show.


[Schematic] [Layout]


Volker
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: THAT 1570 1510 and 1512 Input/Output Circuits

Post by JR. »

Are the TL074 buffers needed? You can use a simple resistive pad to drop levels down to 3vp-p. I guess it's nice to isolate the audio sources. If the A/D inputs are multiplexed (they usually are) their S/H capture acquisition can talk back into the inputs, and settling time for the S/H will affect crosstalk/noise floor (jumping between a loud input and quiet one, means the sample cap has to slew to the new level before conversion will be accurate).

If you want to make the TL0's output impedance a little lower, use a small feedback cap across the feedback R. They are not generally low Z opamps open loop. If the audio is coming from any distance you can improve ground noise isolation by adding a couple resistors to make those inverting buffers into 4 resistor differential amps. I guess it depends on the resolution of the A/D, it might not matter if their dynamic range is very limited.

I have never tried to push serial com over that kind of distance. I had issues with SPI over inches when my clock edges were not clean and data would get corrupted, but you should be OK using a dedicated driver.

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: THAT 1570 1510 and 1512 Input/Output Circuits

Post by JR. »

I don't think he needs to go that far,,,he's probably just citing the driver data sheet numbers.

I2C and SPI are generally included in the firmware for popular micro's while i rolled my own SPI once to deal with circuit issues in one design.

When mixing digital poop with high performance audio poop, sometimes the digital noise can leak around through sundry paths. I ended up interleaving my SPI data rate with the A/D sampling interval to keep the noise floor under better control.

But every design is different.

JR
Cancel the "cancel culture", do not support mob hatred.
hazmo
Posts: 19
Joined: Wed Dec 23, 2009 7:31 am
Location: Germany

Re: THAT 1570 1510 and 1512 Input/Output Circuits

Post by hazmo »

The cited 50m are just what is doable from my experience and are a reasonable target I think, for 5m
I probably wouldn't build a remote preamp. I've seen I²C with up to 200m. The protocol doesn't
specify any maximum distance, only the data rate has to be adjusted. Also the bus driver datasheets
only state guarenteed distances under worst cases. As JR says, I²C is implemented in hardware
already, RS485 requires a bit more effort. I'll see how it works out.


Re the buffers: the ADC asks for less than 10k source impedance for full speed, so a high-ohmic
resistive divider won't do. I'll probably go differentially (and also put the buffer on the preamp board),
since the signal in the preamp is differential all the time. Would the effect of the single ended loading
be negligible or should it be avoided?


Volker
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: THAT 1570 1510 and 1512 Input/Output Circuits

Post by JR. »

Sorry I haven't been paying attention to what you are capturing the audio for. A/D specs inside micros have some specific gotchas. The micro family I use, provides info on internal parasitics and impedances. Basically your source impedance, must charge up (or discharge) the A/Ds input hold cap, so this source impedance can impact raw acquisition speed if too high impedance. Namely if you are multiplexing between multiple inputs, you must ramp the input hold cap from the last input's voltage to the new input's voltage, before you can even start the conversion. Even with zero source impedance, you will have a finite internal impedance in series with the internal hold cap, due to the multiplex switch on resistance that you can never do better than. I had some salesman puke trying to sell me on adding his low voltage opamps, but from inspection, a passive pad, with a cap to ground can be fast enough to not obscure the audio but low enough impedance (thanks to parallel cap) to suppress multiplexing artifacts and support sampling speed.

But I can't speak for your hardware, only my personal experience. Note: This input acquisition time is different than the A/Ds settling time for rated resolution which is a different mechanism related to the settling time of the individual sequential SAR approximations. My micro A/D section had separate tweaks for both of those and more variables.

In my last meter design where I was multiplexing between 6 inputs, I did a bunch of my own tests, on the working unit to optimize resolution, bandwidth, and get acceptable crosstalk figures. Crosstalk can be a very unprofessional looking flaw in a meter, where a hot channel shows up in the following sampled channel's noise floor. It looks like a sloppy layout, but it is an artifact of input acquisition time window.

have fun... there are any number of ways to skin any cat... I work at skinning them without using extra parts.

JR
Cancel the "cancel culture", do not support mob hatred.
MIQ
Posts: 3
Joined: Tue Aug 03, 2010 4:55 pm

Re: THAT 1570 1510 and 1512 Input/Output Circuits

Post by MIQ »

I was wondering if anyone had any ideas for interfacing electret capsules to the THAT1512 preamp ICs. I've been developing a mic pre for taking acoustic measurements based on the THAT1512 and it works great with P48 powered mics via XLR inputs. We would also like to use electret capsule mics with internal FETs biased with a 10K resistor to a +9VDC supply. The issue I am having is trying to determine the best way to connect the unbalanced signal from the high impedance FET grounded source stage to the low impedance input of the THAT1512 circuit. The trade off I'm struggling with is between the extra noise of presenting the input of the 1512 a high impedance (~5-10k) looking back at the FET stage vs the extra complexity and cost of a low noise impedance converter (5532 follower?). This high impedance at the THAT1512 input would only be switched in when the electrets are being used and won't affect the balanced XLR input configuration. The mics are relatively low noise for electrets (-108dBA ==> 4.5uV @ 20-20k BW - calculated). I tried to measure the EIN of the THAT1512 with 10K between the + and - inputs with the - input connected to GND. It looked like an EIN of 2.4uV (20-20k BW) without any enclosure around my breadboard. Seems like the mics self noise will dominate this but is there some gotcha I'm missing? When I wire a 5532 follower in between the 10k and the THAT1512 and bring the impedance down to 680 Ohms at the 1512's input, the noise gets worse. I'm thinking my bigger unbalanced input loop is picking up more stuff on my bench. Anyone have any experience trying to use electrets with the 1512?

-Mike
Post Reply