3458A upgrade questions

[essele]

I find myself with a lot less cash, but with a nice 3458A to play with after finally biting the bullet.

It appears to be an early 90’s option 002 model and I’m just running through some SN18A stability checks for a few weeks before I fully commit (thanks to some good terms with the non-ebay seller.) But while I wait I’m obviously thinking about upgrades...

The Dallas ram’s are all ‘91 dated, so I’m going to need to sort those out ... I’m keen to install FRAM to replace these following the examples in this forum. I’m nervous about desoldering the old devices ... any advice on approach here would be appreciated!

The firmware is currently v6 (also dated ‘91) so I want to upgrade that ... rather than erasing the EPROMs I’m thinking about just installing some newly written OTP EPROMs (for example AT27C512) ... is there any reason why this wouldn’t work? They seem to be a low cost and readily available option??

The Option-002 reference (03458-66519) looks to be identical to the standard model, so I assume the improved stability is through selection only, therefore would this also benefit from the oven temperature reduction? I assume so, but I can’t find any discussion about this anywhere.

It’s really exciting ... I was super pleased with my Prema 6048, but this seems like it’s in a completely different league!

[TiN]

Desolder carefully. It's no big deal , no sensitive stuff on digital board.
OTP EPROMs will work fine.
All references are exactly same hardware-wise, so it's no difference. Opt 002 or not, doesn't matter really in practical use.

[TheSteve]

If you care about the existing cal data you might want to try backing up the existing NVRAM data via GPIB before desoldering. I had no problems reading the data out of mine after desoldering but others have.
The boards are pretty good quality. Add a little fresh solder to each point before desoldering, any decent vacuum desoldering tool will make short work of it

[ap]

The good thing is your ref still has the PN sticker. So formally you can, without a history, apply the 002 spec (still of course, your meter could have the ref drift issue, so powering it for a while should be done anyhow if you have no history of it). Older references by now have reached the 002 drift rates if the meter was properly used, but you would have to find out.
Reducing the oven temperature helps reduce the drift also for the 002 ref.

[Dr. Frank]

Hello essele,
welcome in the 3458A club!
As the others already have said, it's no problem to replace these nvRAMs, especially the 2k CAL RAM, if you first save its data via GPIB, and then later transfer these cal constants to the fresh nvRAM.. I assume you have an appropriate programmer available.
Desoldering is very easy, as the pins are not sitting tight in their holes. If you use  the same DALLAS chips, precision sockets are recommended, maybe also for the FRAM solution, if something goes wrong with the adapter.
It's a miracle, that these 30 year old data are not yet corrupted, or do you get error messages?

I see no problem to use OTPs instead of EPROMs, as FW 8 or 9 are the latest versions, even in the new Black Edition.

Your 2ppm/y option means, that the LTZ1000A already drifts that little, that a reduction of the oven temperature will not give any added value.
In contrary, you would initially have to do a re-calibration.

The modus operandi for the specified 8, 4 or 2ppm/year DCV drift is not described in the manual. So what is the 'proper use' ?
That oven temperature reduction makes most sense if the instrument is running 24/365. 
If the instrument is turned on for a small period of time / year only, the drift should anyhow be very low, as a 'cold' LTZ1000 has nearly zero drift.

I wish you, that SN18A does not apply for your instrument, neither for the reference, nor for the U180.

Frank
 

[essele]

Thanks to all for the comments and answers ... very helpful.

I've decided not to lower the oven temperature ... seems like a potentially unnecessary step at this point in time.

I can get prepared with the OTP EPROMS, so I'll probably order those soon ... I do have a couple of different programmers, so this shouldn't be a problem (let's hope I don't come to regret the OTP aspect!)

I've been looking at the FRAM options and for the CAL RAM I really don't like the 28pin adapter hanging out (it will bug me knowing that it's in there), so I will either look at programming the chip separately and build a simple DIP24 adapter, or I'll see if I can come up with another solution. Maybe I just go with new Dallas chips and forget about it for a few years (although that seems like a less interesting solution!)

It is stunning that these 1991 devices are still working ... I don't get any errors, but surely I can't expect them to go on much longer (I will take a GPIB dump as soon as I can.)

Sadly all of the actual implementation will have to wait until stability is known.

[TheSteve]

Reprogramming the factory EPROM's is also an option. I did that on both of my units. Once I dumped them there was no reason to keep them around with the ancient firmware in them(one of mine had version 2). I just lifted one end of the original label, erased and then reprogrammed.

[essele]

This has progressed quite well ...

- Replaced fan, still quite noisy, added some thin rubber grommets to the fixings which helps slightly.
- Replaced mains filter, fiddly job ... dropped a washer into the case a couple of times, thankfully recovered.
- Programmed OTP EPROMS with V9, all works well.
- Fitted sockets for the NVRAM and CAL RAM - not as scary as I thought!
- Made some adapter boards for the CAL FRAM, was able to program the FRAM using a SOIC to DIP adapter and a GQ-4x4 before soldering.
- Made some adapter boards for the other FRAM (could have just bought these, but why not!)
- Replaced all the caps on the inguard PSU (A4)

Annoyingly I made a mistake with my cap order for the A6 board, so I've got to wait before I can tackle them.

Programming the FRAM was interesting, it's supported with the GQ-4x4 programmer and the ZIF SOIC to DIP adapter worked really well. However I did see some strange behaviour ... I could program it with all zero's or all one's or any single value and it all programmed fine. When I tried an 8K image (four copies of the 2K image) it seemed to write some bytes incorrectly (even in the first 2k), yet if I programmed just the 2K image (followed by zeros) it all worked fine. Very strange -- but not a problem for this!

I've checked an ACAL and can see new values written successfully, and I've programmed a couple of the keys and that all seems to work fine also.

Thanks again for all the help and advice.

[Dr. Frank]

Looks very nice & tidy.
Could you please publish the CAD data for these boards?

THX
Frank

[essele]

I've attached a zip file with the KiCad files for each board, and it also contains two further sub-zips for the gerbers and drill files ... these are exactly what I uploaded to JLCPCB, so they should be ok.

For future reference, on the CAL one, I've left the unused pins floating ... with hindsight I should have grounded them, that way you could have used the larger fram devices if needed.

[essele]

The remaining work is now complete ... I've replaced the caps on the A6 board, and tightened the mains filter since it's much easier to get to once the A6 board is out.

I also re-replaced the fan .. I had originally ordered an EBM Papst 612NLE (which looked to be pretty much identical to the original), the datasheet suggests 16dB(A) but it definitely seemed louder than that to my very untrained ear, and sounded identical to the (presumably 30 year old) one I took out.

I tried using some rubber washers on the mounts, and then a full 60mm rubber gasket ... it helped a bit, but I think there was sufficient vibration from the fan that it reverberated through the case (probably need different fixings as well to avoid this.)

So I've now replaced it with an AcoustiFan 6025B which has a slightly higher airflow (13.7cfm vs 12.4cfm) and is supposedly noisier at 16.9dB(A), but it's actually much quieter (both outside and when mounted inside the case) ... current draw was exactly the same. I kept the rubber gasket in place.

I'm not sure if I received a bad unit or whether there are different ways they measure noise!

The Papst fan has a quoted lifetime of 135,000 hours (15 years) whereas the Acoustifan is only 60,000 hours (6.8 years) ... but I think I'll worry about that in 6 years time.

One question did spring to mind whilst I've been doing this...

There does seem to be a reasonable level of temperature variation inside the case (i.e. a fixed amount above ambient) where my ambient temperature does swing around by two or three degrees, unfortunately it's not a temperature controlled lab.

Would there not be a benefit of installing some fan control to try to keep the inside of the case more constant?

I'm guessing there are problems doing this:

- You'd have to keep it at the higher end of the temp range ... would that impact resistor stability?
- Would the variable airflow be a problem?
- Would the fan control system introduce noise? (I was assuming some kind of linear approach rather than PWM though)
- I assuming it would have limited upper range ... but outside of that it would just revert to fan-on as per normal.

Given it's a simple thing to add, and it's not there by default I'm assuming there's no value in it, but thought it was worth asking!

[MiDi]

Personally, I would not do a fan control mod.
I suspect that changing the airflow will lead to unequal temperature changes in the instrument and different settling times, which could do more harm than use.