Ultra Precision Reference LTZ1000

[MisterDiodes]

Not that it can't be done for just fun, we've never seen a serious attempt at thermal simulation that comes very close to any -real life- LTZ module design.  All in all the LTZ can itself is pretty forgiving but you do run into other issues when trying to run any accurate thermal sim.  Typically you don't read too much into the thermal sim results for any LTZ design.

For instance, the various combinations we deal with:  Are you connecting the LTZ to inner trace or outer? Copper weight? Number layers?? Ground planes??  What PCB material (FR-4, High Stab Rodgers, ??)? Size of board??  Board mounting method to shield box?? Soldermask, and what type?  Conformal coating and what type? Air draft shield around LTZ and what type / material / thermal flow??  Solder mass at each joint?  Ambient temp around LTZ?  Component distances from LTZ and what is their thermal dissipation, component lead length and bend radii, number of bends, emmisivity. PWW or Metal Foils resistor cans or combo?  Distance to board connectors, and what is connected to each board connector, number of connectors, wire gauge solid or stranded??  Ambient temp of shield box?  Overall thermal mass and time lag?  Is the interior of the box purged with CO2 or Nitrogen or Air or ?  What is the LTZ board mounted to and how does that structure act thermally?

And so on.

All of those elements play at least some small role in exactly how every LTZ finds it's unique & stable operating point, but the on board heater does a pretty good job of isolating the LTZ die from the rest of the world.  We've never seen voodoo slots do anything helpful, at least in our tests, and neither does the very successful HP3458a Vref module, for decades.  A (much) more accurate thermal investigation would involve just building some real designs and comparing sim theory vs. real-life reality (remember reality wins every time) - you'd get a better feel for what's going on, and gain more knowledge faster - at least that's what we've seen.

Have fun though!

[zhtoor]

For instance, the various combinations we deal with:  Are you connecting the LTZ to inner trace or outer? Copper weight? Number layers?? Ground planes??  What PCB material (FR-4, High Stab Rodgers, ??)? Size of board??  Board mounting method to shield box?? Soldermask, and what type?  Conformal coating and what type? Air draft shield around LTZ and what type / material / thermal flow??  Solder mass at each joint?  Ambient temp around LTZ?  Component distances from LTZ and what is their thermal dissipation, component lead length and bend radii, number of bends, emmisivity. PWW or Metal Foils resistor cans or combo?  Distance to board connectors, and what is connected to each board connector, number of connectors, wire gauge solid or stranded??  Ambient temp of shield box?  Overall thermal mass and time lag?  Is the interior of the box purged with CO2 or Nitrogen or Air or ?  What is the LTZ board mounted to and how does that structure act thermally?

hello,

that essentially means the reference system as a *WHOLE* has to be simulated from
a thermal point of view keeping in mind thermal sources and sinks reasonably accurately,
which i suspect should be pretty difficult (albeit possible).

regards.

[3roomlab]

yea im jus fiddling around femm for fun, but it is very helpful cos even a FLIR wont see thru thermals like this.

the trace are connected to pin (see pic 11), on 1 side. for edge box mating, all materials meet the "box". i also tried to simulate solder, but the effect is too small to see near the LTZ, however, a large blob outside have a interesting effect (pic 101)
in 2.gif, i exaggerated the sphere foam to 100x less conductive to magnify the effect of the heat going thru the PCB.

i also tried a simulation with long kovar pins going sideways. comparatively, copper trace need to be very small to match the same thermal resistance of KOVAR, by the time i am able to get a 3C difference using 10mW as heat source on 1 pin (300C/W?), the copper ends up being 11mil wide @ 16m-ohm x360mil long. resistance wise, this copper trace is about 16mohm, the kovar is about 20mohm. so in terms of the experiment andreas did regarding seating height, it is low seating with joint temperature controlled by LTZ or extremely well insulated joints further out (over 1000C/w). the blob of solder also have a interesting "heat capacitor/buffer effect" see 102.gif.

then i try a combination of slot and sink (see 100.gif), creating a deliberate temperature "short circuit to ground", the effect is also very interesting. 100.gif is 4 dots, 1 of which have a copper trace running to the left edge leaking heat into FR4 around it. it passes thru slots and another copper sinker

(TBH, i have no idea what im doing fiddling with these)

[Vtile]

Now when you are at it, what about those connection whiskers hanging between the wafer and the legs. Cool is that some dedicated simulation program or DIY?

[3roomlab]

Now when you are at it, what about those connection whiskers hanging between the wafer and the legs. Cool is that some dedicated simulation program or DIY?

i got the idea of the side section from a pic here (photo by eevblog member chuckb)

i assumed the whiskers dont do much.

the program is by these ppl
http://www.femm.info/wiki/DavidMeeker
free to use
http://www.femm.info/wiki/HomePage

edit : after some more thinking about many things misterdiodes said and the results posted by our hardworking andreas with his many many many many versions of LTZ, it seems to me that if an "uncontrolled" power source drives the LTZ, the fluctuations in power consumption = fluctuations in ppm = small thermal variations, what if the LTZ is supplied with accurate CV and CC = constant power = a fixed thermal leakage? or it has already been done in such a way?

[MisterDiodes]

The LTZ -when the heater is working and holding the die temperature steady - is not going to respond much to reasonable (small) supply voltage variation.  The Vref output voltage is itself overwhelmingly set by the zener junction and it's transistor driving the '1013 current source - and that Vref voltage will be different for every individual die, but very stable.  Small  supply voltage changes and even a Voffset drift of the '1013 current driver do not have any major effect on Vref.

That's the beauty of the circuit.  Elegantly simple, pristine analog circuit yet very, very effective and stable.  Every element you need to be ultra-stable is already in that LTZ can, and out of your control for the most part.  The rest of the components are relatively forgiving, as long as you watch out for air drafts, EMI problems, etc.

Sometimes we use a linear voltage pre-regulator just ahead of LTZ as an additional noise suppression technique for our particular situation, but that isn't really needed for a lot of applications.

That's what I meant in my post above:  The on-die heater does a very good job of isolating the die from the rest of the world AS LONG as it can regulate temperature properly - that is the foundation on which everything else is built.  It is a heater only, not a cooler.  Therefore you have to keep the die a bit warmer than the variable temperature surroundings, and there -must- be some thermal flow -out- of the LTZ, at least in a controlled way.  As long as you understand that, then you understand why you can't over-insulate the LTZ or run it too cold - which means too close to ambient temperature (heater ratio resistors too low).  The die must must be held at some higher headroom temperature difference between all of its surroundings.  That means you look at your application requirements and set your heater resistor ratio based on the expected high temperature of the local environment.

In other words:  Let the on-die heater system do it's job however it needs to, use good fundamental design rules (compact board, star reference points, keep an eye on trace voltage drops, thermals, shielding, etc.) and then (most) everything falls into place to give you a very stable Vref output.

[Andreas]

i got the idea of the side section from a pic here (photo by eevblog member chuckb)

now you can improve your simulation model according to the real world issues:

- different length and number of bond wires
  (the chip is most likely not symmetrically cooled by the bond wires.)

- chip not exactly centered

- different temperatures of the on-chip thermo-couples
  Aluminium to Gold bond wire to Kovar Pin

the only thing that can be influenced is trying to keep the pin to PCB connections (Kovar to lead to copper) at the same temperature. But I am not really shure what is better cutouts or not. Long or short leads. We had a long discussion with the LM399 thermal photos. My impression was that a certain thermal mass is necessary to keep the pins all at the same temperature.
(And perhaps some slots at a 1-2 cm distance to keep other heat sources away).

  We've never seen voodoo slots do anything helpful, at least in our tests, and neither does the very successful HP3458a Vref module, for decades.

Mhm,
- the venerable Datron calibrators use slots (even with flex cirquit) around the LTZ in its 4000 series.
- Agilent (HP) now also uses slots in the 34470A reference cirquit. (have they learned something during last 30 years?)

The LTZ -when the heater is working and holding the die temperature steady - is not going to respond much to reasonable (small) supply voltage variation. 

Ok for small variations in the 100mV range this plays no role.
But -0.4ppm/V variation is something where I do not want to have a unstabilized power supply.
https://www.eevblog.com/forum/metrology/ultra-precision-reference-ltz1000/msg842662/#msg842662

And even with a linear regulator you have to take care that it is not too close at the LTZ1000 if you want to have less than 0.1 ppm/V. (or you can use isolation slots between voltage regulator and LTZ to improve the PSRR).

https://www.eevblog.com/forum/metrology/ultra-precision-reference-ltz1000/msg846835/#msg846835
https://www.eevblog.com/forum/metrology/ultra-precision-reference-ltz1000/msg935102/#msg935102

With best regards

Andreas

[MisterDiodes]

In our experience, voodoo slots have never done anything helpful except when you want to build a sensitive vibration detector out of an LTZ  .  There might be applications where a slot is required to suit some design.    I would never add them just for looks though.  Never seen slots on a 3458a Vref., and on an early test where we made the same board design with and without slots (5 boards each) we could see slots did nothing positive, at least for what we are doing.  Except the slotted board had vibration problems unless we padded the mounting.

We use slots only very, very carefully and only when we know they are required - If you get something flapping around loose in the breeze on a production circuit you'll always find out about it - in a bad way.  Like when you find out your circuit is picking up the lumbering freight train down the road and you can tell it has loaded cars or not....<yes, that has happened>  So if you add the slots do it only if you -know- for sure you need them - and then I would look at a vibration isolation mechanical mounting system.

Notice in my previous post I said "small" power supply deviations were allowed.  You don't want the LTZ's power source dancing around too much, and use a pre-regulator as required.

Yes, I should have been more clear: We don't put the linear regulators on the same board as the LTZ...the LTZ is always on it's own module for what we do, and that lets us do burn-in on tester jigs ahead of time while the rest of the system undergoes separate testing.

[TiN]

I wouldn't use 34470A design as a very good example of LTZ design for purpose of long-term stability/reference standard role.

[pigrew]

Do slots help reduce die strain (due to TCE mismatches)?

Or is the die attach somewhat flexible reducing strain effects?

For example, the LTC6655 suggests cutouts around its package.

[Micke]

MisterDiodes,

I really enjoy your posts and have learned a lot, but I have a question... 
I understand you burn in the LTZ1000 assemblies for like a year, but do you do any temperature cycling, or are they just powered in room temperature?

I have built 3 LTZ1000:
2 LTZ1000A  Burn in time so far almost 5000h
1 LTZ1000   Burn in time 3000h, on this I tried some temperature chock treatment, with just the LTZ1000 soldered on the PCB, I heated the PCB to +125C and put it in the freezer for a while and then cycled this like 5 times. But I have read that maybe this was to extreme, I should have instead slowly narrowed down the temperature span.
But anyhow, so far the LTZ1000 (non-A) shows less power cycling hysteresis than the LTZ1000A, I read it should be the opposite, LTZ1000A should have less power cycling hysteresis. Could this be that the LTZ1000A just need even more burn in time?

[d-smes]

I wouldn't use 34470A design as a very good example of LTZ design for purpose of long-term stability/reference standard role.

I value your opinion.  Can you please explain why?

[Kleinstein]

The LTC6655 and similar SMD parts are much more sensitive to board stress. For this reason they might need some slots or even a U form cut. The To-99 if the LTZ is rather insensitive to board stress. The metal case and open air bond wires keeps stress away from the chip itself - especially if soldered with a little spacing to the board.

I the reference circuit is on a small board by itself, there will be not much stress anyway - so no real need for extra stress reduction - just make sure not to have a poor connection to the main board part that causes massive stress and temperature gradients.

For the temperature gradient the copper layout is more important than cuts in the FR4 - much of the heat flow is through the copper.

[MisterDiodes]

A couple of thoughts below, and remember this comes from experience building profitable circuits in a production environment, and where customers pay for reliability and 24/7/365 operation, OR where the device might go into shutdown for an hour, a day or a month or more.  For us, failure in the field is NOT an option (at least in spirit), and if there is a field failure we take that very seriously and try to improve the product so that particular product failure -never- happens again.  A hobbyist may or may not choose to do it this way depending on what the application goal is.  I have to just keep it to the highlights here, I can't give you an exact recipe.  Use the information at your own risk, since your situation may call for a different procedure.

RE: Burn in procedure:  Yes there are many temperature cycles that go into burn in, which is a combination of a long steady run, and then a period of thermal cycles, etc.  We throw everything at the LTZ and it had better work within spec after every trip thru the thermal chamber.  This is usually a good way to weed out bad performers on other areas of the board too.

You have to realize that if your product is shipped air freight in an unheated airplane cargo area, it's going to get very cold.  Or if it goes to somewhere like a hot customs inspection area and sits for a week its going to get very hot.  The customer doesn't care - they need it to work out of the box, and we provide guidelines for a minimum stabilization time before guaranteed specs are met after shipping.  Hint: it's not really the extreme temperatures that causes real issues, it's the -rate of change- of temperature that can cause surprises on your Vref., so that is something we test for on every device to meet customer design requirements.  Sometimes you see something break with a small 30°C shock test, and it might have worked fine before over a slow 100°C ramp.  Everything around the LTZ has to remain in a working condition after every thermal cycle, no matter what we throw at it in our test design.  Our test will include the customer requirements plus another safety factor to make -absolutely sure- it works to specification at customer site - with a comfortable margin for error.

We test at the whole range -65°C to 150°C for storage, and that can be over a day or some weeks, and then there is the actual operating temp range which varies by customer design requirement.  We never use the non A version, only the LTZ1000A (same as most 3458a's, which we find overall is more stable over the long term, at least in our tests) and yes there is a small power cycle hysteresis but that should be well under 0.5ppm on a stable LTZ, and most of the time we need to hit a 2ppm or 4ppm stable operational window - so we're not trying to replicate a Fluke 732.

We do develop a stabilization time procedure for the customer so they know when it should be back in spec after a power down of say 15 minutes, 1hr, 12hr, 24hr and 30+ days.  It should recover typically within a couple hours or a day or two, sometimes longer...but even then it should stay within the guaranteed 24hr drift rate spec.  Like any precision equipment if it has been powered off for a long time, you expect to power it on and let lit stabilize for some time before it goes back to work and settles back to "in spec" condition.

It just depends on the design requirements, and usually each customer has a very specific need.  Our job is to deliver customer satisfaction - because happy customers are the very best variety.

RE: Keysight 34470a as design inspiration - That's not really a metrology grade instrument.  The Keysight sales guy brought in an early model to demo, and the stupid thing drifted more in 4 hours than an old reliable 3456a drifts in a month.  It could be better now, but look at the specs - be careful when they start listing accuracy in % rather than PPM and compare 24V 10V range accuracy to a 3458a...and let me know which one is much more stable.  Plus those danged PLayskool child-safe recessed banana plugs tells you right there they aren't paying attention to low-thermal design goals (I know they are trying to sell to the safety-spec market but that is not the metrology market).  I know the 34470 is not a replacement for the 3458 and is at a price point to match - but the new design techniques don't inspire anything from a low-ppm tool point of view.  They may have some of those problems worked out but we sent the salesman out the door with his stuff...he didn't sell us on any of newer, cheaper DMMs.  Our fleet tends to stay at the 3456a / 3458a style for lower drift.  In the long term the 3458a's don't cost much to own (for the accuracy provided) since they are usually very reliable - and if you get it under the extended warranty.

RE: Slots on LTC6655... That package is SMT, and is prone to every circuit board stress / flex problem known - which really shows up on the trimmed die resistors as increased noise.  Not a low ppm-style part - that's like a toy compared to LTZ, but not every design needs an LTZ either. Sometimes even slots aren't enough for that package, but use what the design calls for, and in this case slots help and are recommended.  But comparing a '6655 package to an LTZ is like comparing apples and oranges - the LTZ is much more insulated from PCB mechanical stresses.

The other thing to watch out for with slots - be careful that adding slots doesn't increase trace length so much that now you've increased EMI / current loop antenna effects.  If you try to help too much in one area you wind up with decreased performance in another area.

Have fun!

PS EDIT:  Also look at something like a Datron 4808 calibrator 10V accuracy spec in both long and short term (apparently with a slotted LTZ, I haven't seen this in person) and tell me if that's really different from a standard 3458a...and then compare it to a High Stab 3458a with 02 option (exact same board as any 3458a still no slots, just a rather well performing selected LTZ).  Is the calibrator LTZ Vref design really more stable overall than what is really the basic LTZ datasheet circuit in the 3458a?

[3roomlab]

now you can improve your simulation model according to the real world issues:
- different length and number of bond wires
  (the chip is most likely not symmetrically cooled by the bond wires.)
- chip not exactly centered

after a few more fiddly diddly
here is what the FEMM thinks of the 3 bond wires vs 1 bond wire. ~ approx 38uW diff (0.05C?). pic TT6/7
uniformity of connection to LTZ island TT8.gif, the bottom side lacking copper fingers = higher temperature. diff in top side island and bottom side island is approx 1C

**update
i corrected the W/mK error for plan view. hard rubber mimic nearly FR4 0.8mm thick, lead liquid actually mimics very close to a copper pour back planed 0.8mm FR4. all 1oz.
now with this, the thermal leak presented by a trace looks more corrected. so the simulation for using small widths connected to the +10C source (LTZ) and using it as "resistor" to keep the opposing temperatures "far" from each other. (pic P1) the center ground plane appears to keep uniformity very well, the right most temperature point is simply a +10C point on bare FR4 for show
as seen on the LTZ model 3 traces, the top trace 10mil by FEMM contour calculation, the leak is 26uW. left trace is a 6mil 400mil long to a fatter 50mil, leak is 24uW. bottom trace is 50mil straight from the +10C pin, leak is 40uW. so as we can see, the 6mil is gving good resistance. of course i think to consider uW leak level is overkill, but i think by varying the length it can help to keep all the pin temperatures uniform, which will need extreme fine temperature measurement. but on the other hand, after introducing an outer ring of thermal impedance, by introducing a small fat ground plane inside the LTZ island under or top side, without varying the trace lengths, the centre island can have its own uniform temperature by self regulation (P2 pic)
** addition
by applying the new W/mK to side view. and go with a tiny copper spreader and use the hot pin as a "preheater". we might force the cooler pins to within 0.1C (in theory). or even crazier, put the LTZ on a tiny ALU sub-board for LED. on ALU, the simulation show the temp diff is now under 0.1C for sure (in theory).
(now what improvements would we expect if these are applied to LM399?)[/size]


new update **
a little imp sprung to mind after thinking about misterdiodes mention of heaters. so i modified the simulation to have some 0603 resistors heat the legs. assuming you can heat at a good rate (simulation suggests 10mW per vicinity of a pin), it appears it could be possible to bring all the pins to 0.02C of tolerance from each other.
simulated are A001.gif / A003.gif
only topside heater and or top and bottom heated.

and A002.gif is an overkill heater layout of 16x 0603 heaters in 4 series strings of 4. assuming 12V supply and total heating of 80mW, each 0603 (5mW) needs to be about 1.8k. or we could use all 1k and operate a variable CV to regulate the actual mW output. (i noticed i made an error in 1 of the materials, but still the model still allows us to see the behavior of the temperature balance)
on a top only possibility, and to create a larger thermal mass, it is likely possible to stack on larger resistor sizes @ 0805

[MisterDiodes]

For anyone who's forgotten, Dave Jones did a teardown of the 34470a here:



About 14:15 in you'll see the LTZ Vref board construction, and notice they are going for a much lower price point with the SMT components.  Nothing but penny resistors in there, and SMT caps.  This board will be very prone to mechanical vibration pickup, but they do have that somewhat isolating plastic clip that might help (or make things worse).  There are slots around the LTZ and a slot to isolate the heat from the heater transistor.  Still 15k over 1k heater ratio.

At 16:00 you can see the removed bypass cap (or whatever they were trying to do that didn't pan out - or maybe test points) under the LTZ.

But the accuracy / stability of the entire meter suffers due in part to this design and has relatively fairly low 24hr accuracy.  This is a good example of what happens when designers go the cheap route - this is one reason why a 3458a runs circles around the cheaper DMM's (for low ppm stability / accuracy spec).

ON THE OTHER hand it also shows how forgiving the LTZ is even when used in a somewhat lower price point application.  If you consider about $3900 cheap for this 7.5 digit DMM.  Even at that price we found a better value and better accuracy in a used 8.5 digit 3458a, but that's what we need.  Other users will have different needs.

It would be interesting to see if anyone has made a better-quality LTZ board for this meter, and did that help overall performance?

[kj7e]

I think the greater limiting factor with the 34470 is the tempco from all the surrounding components and not the reference board itself.  I suspect, the LTZ is simply overkill in the 34470 and the meter as a whole just cant gain a significant benefit from it, other than perhaps long term drift.

[TiN]

Dr.Frank did good detailed summary on this design in Dave's video discussion thread. Somebody at PC can surely find link to it and post here. SMD esistors could be still VPG foils , but I never see VPGs BMF with such style, they are usually light-blue or encapsulated in epoxy. Either way SMT assembly not great for stress and vibration. Also those MLCC's there are great microphones, and vibration sensors from meter's fan. Piezoelectricity is reason why I used only film caps on my reference, even though they cost few $ a piece.

[Edwin G. Pettis]

The resistors on the board in question are film, not foil, for one thing if they were going to foil they would have used the larger non-SMT package, this looks like a cost cutting design in just about every way.  The noise is likely to be higher than if they had used foil or PWW resistors and you are supposed to use film capacitors not low cost crappy SMT ceramics, no this is not a high performance Vref board compared to the 3458A's or other similar versions, this design barely manages to get near the 'claimed' 7.5 digit performance, the specs are not great by any comparison.  An old 3456A would give this bird a run for its money for stability and long term accuracy.  The newer DVMs are of questionable design compared to the earlier generations.

[lowimpedance]

PS EDIT:  Also look at something like a Datron 4808 calibrator 10V accuracy spec in both long and short term (apparently with a slotted LTZ, I haven't seen this in person)

It indeed has slots, see this old post for a few pictures of it.
https://www.eevblog.com/forum/testgear/datron-voltage-reference-pcb-module-pics/msg813331/#msg813331

[MisterDiodes]

...and after all the time spent on those floppy slots, for no net gain on the instrument daily / yearly drift rate or noise over the slotless design.    The advantage on the 3458a compact module style is the small board size - If the LTZ is mounted on a larger board then stress reduction slots might add some benefit as long as trace length isn't compromised and as long as weird draft flows don't develop, etc.  OR if you have to deal with a heat source coming in from a certain direction, then of course you need to tweak the design as required.

The other weak area on the 34470a LTZ module design is that cheap header connector - if those header sockets are allowed to freely float like that they will tend to wear and go intermittent as the gold wears away on the contact pins - we've seen failures like that especially after a header connector gets loose while traveling on FedEx somewhere, and then both the header pins and sockets need replacement.

I agree with Edwin:  There is not a lot on the new cheap(er) meters that inspires confidence in the design expertise of the product engineering team at Keysight (or elsewhere) these days - even on a non-metrology grade instrument.  Of course it is not a replacement for a 3458a and never promised to be, but it's a shame they cut corners to that extent on a meter that still sells for almost $4k.  If it was a cheaper meter yes it's understandable.

Of course you get faster readings and a whiz-bang display in a smaller, lighter footprint, but at the end of the day we still need low drift and good stability - for what we do at least.  These meters tend to fall way short compared to what you can buy on the used market for about the same cost that works much better if you need better stability.  You also get good quality low thermal binding posts on the better quality meters, and if you're doing low ppm measures: that's an important consideration.

It just depends on what you need.

[Edwin G. Pettis]

Another glaring rookie goof up in the 34470A video, the designers placed the LTZ module right over the clock oscillator on the main board, are they serious?   It is sitting right on top of the crystal, gee what could go wrong there?

[technogeeky]

Another glaring rookie goof up in the 34470A video, the designers placed the LTZ module right over the clock oscillator on the main board, are they serious?   It is sitting right on top of the crystal, gee what could go wrong there?

You're probably going to feel a bit silly, but: that's not a crystal. It's a resistor.

From the pictures here, we can see that that component is populated (but not labelled on the silkscreen) on the 34470A (board 34465-xxxx):



And it is silkscreened but unpopulated on the other board (board 34460-xxxx):



I only investigated because it seemed so boneheaded to put a crystal (usually controlling a digital system) right next to an analog reference, so I was curious to see it. Turns out, nope! Just a precision resistor.

[Edwin G. Pettis]

Interesting, in the video Dave identifies that can as a crystal, you can't read what is written on it unfortunately.  Unless it is identified on the schematic as a resistor I can only go by Dave's commentary that it is a crystal.  So we have a bit of a mystery here, what is it a resistor or a crystal?

[technogeeky]

Interesting, in the video Dave identifies that can as a crystal, you can't read what is written on it unfortunately.  Unless it is identified on the schematic as a resistor I can only go by Dave's commentary that it is a crystal.  So we have a bit of a mystery here, what is it a resistor or a crystal?

It's a resistor. That's why I linked the photo of the 2nd board: it's R324 (unpopulated on that board). I can't read what is on the part, but text wise it looks very much like a Vishay metallized (oil filled?) resistor. But we can be 100%certain that it's the same part/location as on the unpopulated board.

Also, a crystal in an instrument like this would probably be a 4-pin crystal. (And it is: it's a tiny 4-pin one, 106.25 MHz, to the bottom-left of the Lattice chip).

So Dave just made a mistake in speaking, no biggie.