[time-nuts] Fury Interface Board: How about TI OPA277?

[kevin-usenet at horizon.com]

>> One trick I use to get better tracking is to build ratios out of multiples
>> of a single resistor value.
  
> Can you guarantee this works every time?

Define "guarantee".  For a reasonable risk ratio, yes.  Resistor
manufacturers don't go out of their way to package resistors from
different batches on the same reel, so you if you use adjacent parts from
the reel, the odds are overwhelming that you'll get a uniform material
composition, which leads to a consistent tempco.

That's certainly good enough for me to build a product out of
the parts before doing the temperature testing.  (It's not
like parts always meet guaranteed specs, either.  Remember the
Signetics 25120 Sale price vs. AQL curve.)

If you want it pre-tested, get a resistor network.  Either custom, or
again get multiple copies of the same resistance and connect them in
the ratios you want.

For more on the subjet of resistor tempco tracking, including all
of the above suggestions echoed by a moderately experienced
engineer, see "Debugging Analog Circuits" by Robert A. Pease.
See chapter 2, page 28.

> LT1085 regulators are probably too noisy even when configured correctly.
> The output voltage tempco is probably a little high also.
> OCXO requires a supply voltage with better than 0.01% stability for time
> intervals less than the discipling loop response time.

Huh?  Where do you get that from?  Dont they internally subregulate?
The raw power is used only for the oven... oh.  Yes, and power is V^2/R,
so epsilon change in voltage causes 2*epsilon change in oven power.

Which boils down to a statement that the OCXO should internally
subregulate.  (With possibly a hack to bypass the regulator during
warm-up.)  Does anyone know how this works?

>> - Wouldn't it make more sense to configure U1A and U1B as simple
>>   non-inverting amplifiers, that will present a high-impedance load to
>>   EFC_input, thereby avoiding the need for all the J8-J10 and J21-J23
>>   jumpers?  You need a 3.3V reference for the U1A circuit rather than
>>   1.5V, but that's easy enough to make out of a TL431.
  
> Not really practical if one needs to compensate for small voltage
> differences between the Fury EFC gnd and the OCXO EFC gnd,
> However with a suitable star ground and independent supplies for the
> Fury and the interface board this should not be an issue.

Ah, I didn't realize that was an issue.  That makes life trickier.

> There is little point in using a low drift opamp with a TL431 reference.

Good point; I just mentioned the TL431 because it was already there.
FWIW, the LT1019 is excellent.

And once you have one precision reference on board, it and a few pass
transistors can give you a precision reference for your supplies.


> Personally, I like the idea of feeding the EFC to a high impedance
> amplifier.  However, I am afraid that some would object to having the
> EFC go through an op-amp. It will add noise and distortion.

It's going through an op-amp on the Fury board anyway.  If we have any
specs on that, we can try to match it rather than overdesigning.

> I was thinking about a instrumentation amplifier, like the Burr
> Brown INA103.  It will provide complete isolation for the EFC signal.

There's no need for a high-impedance input for the reference voltage or
input ground, only the input EFC, but it would certainly work.

However, the INA103 has, again, a rather high offset tempco
(1 + 20/G uV/C) at low gains.  This is because in-amps are generally
designed for use at high gains, and so the second stage isn't optimized
for low noise.

I like the INA337, but it doesn't have the power supply range.
The data sheet mentions some other parts... INA114?

I'm a little worried about the gain resistor tempco for single-resistor
solutions.  (Which is why I'm fond of the INA337.)  But how much of the
EFC range is actually used during normal operation?  Does it have to
be 1e-13 stable at the extremes?

> As far as (J8-J10) only one would be connected at a time. I would not
> be feeding all 3 at the same time.

I was just thinking, if they were high impedance, there would be
no need to ever disconnect them.

(Another question is whether it's reasonable to use the same circuit
for the 0-10V and +/-10V options, and just install different components.)

> It has been made clear to me that the +24V(oven supply), +12V(oven
> supply) and +/-12V(EFC interface) supplies must be isolated from each
> other.
>
> The problem is that the oven is very noisy and will corrupt the ground
> for everyone else.
>
> This needs some thinking.

H'm... and if you really want the full +/-10V range, fitting a regulator
and op-amp into the 2V of available headroom requires an LDO and a
rail-to-rail output op-amp.

> The brute force approach is to get 3 separate supplies. However, I
> don't think I need to go that far.
>
> Any ideas here are most welcome.

Do you have any idea what the external power supply will be?
It could be taken into account.

It's noise coming OUT of the oven supply that matters, so subregulating
it doesn't help; muting the current spikes requires energy storage
(L and C).  Outside the pi filters, take the EFC supply, drop it
through an LDO, and watch the PSRR specs on the EFC components.

(Remember, the OCXO is not *that* sensitive to its own oven hash,
or it wouldn't work in the real world.)

I often use the little things called "wound beads" or wideband
chokes", e.g. http://www.ctparts.com/widebandchokes.asp, for generic
supply filtering, but you may want a little more L in your pi filter.