Re: [volt-nuts] EU3458A - what appears to be a 3458A for sales in the European Union only
In message <7862525b-4879-0c7c-c13a-9a6c1f4a3...@freenet.de>, "Dr. Frank" write s: Up to then, it was easy to buy and insert four 62256 SRAMs and get the 01 option for maybe 30€. I've done that directly in 1989, for my first 3458A @ university, later in 2009 for my own hp3458A (2001 vintage), and these RAMs are still available today, if you have an elder unit. And now comes the pointed question: How many of you have actually ever used that extra memory ? As far as I can tell, most people simply want to check as many "Opt" boxes as possible, just on general collectors principle, they never actually use the extra memory for anything. So please spare me the tears over KeySight charging money to cover their costs and make some profit on that PAL chip... Poul-Henning, that's a really good remark ..a nyhow I needed that in 1989 at first, when I used the 3458A for digitizing @ 100kHz / 16bit, at first for capturing 256 sine wave periods from my AC susceptometer. Later, I wrote some DMA code to continuosly transfer the data directly into the sizzling fast 20MHz AT-PC... And just recently, I needed that memory again, to digitize the initial drift of a precision 1kV :10V divider .. as the direct fast transfer would have required special code again. Same story for the 34465A, the digitizing option is useless w/o the 2MB memory option, because the operational system of the DMM is too slow to transfer the 32bit or floating point data directly over the busses, although KS claimed it would be completely compatible to the 34411A (which has in fact 16bit transfer also). So, no crocodile tears necessary, sometimes you really need these collectors options. Frank ___ volt-nuts mailing list -- volt-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts and follow the instructions there.
[volt-nuts] 3458A memory extension
It's still there, 03458-88810 Prgmd Pal-1819-0239 For 001, costs 31.83€. The official upgrade costs 698€. Somebody in another forum explained, that this PAL is all you need. Frank ___ volt-nuts mailing list -- volt-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts and follow the instructions there.
[volt-nuts] EU3458A - what appears to be a 3458A for sales in the European Union only
Hello David, the 3458A is not offered any more by the usual distributors here in Germany, like datatec or Meilhouse. That's due to EU:RoHS directive , i.e. leadfree soldering is required, and the last exemption for 'measuring and displaying devices' ran out as of July 2017. So KS found a way to export that lead containg 3458A directly from the U.S., but I do not remember the special judicial trick for that... only that this very same instrument is now called EU3458A. KS up to now had no chance at all to redesign this instrument to lead-free, because many special components were either terminated already, or were custom specific parts, which never had an updated lead-free equivalent, or were through-hole components with no proper SMD variant. Just look at that badly designed LTZ1000A reference inside the 34470A, that's nearly a 1:1 copy, but with lower-grade SMD chip resistors, instead of the original Vishay bmf through-hole ones. Other examples are the fast comparators EL2018, the A/D hybrid, probably many of the FETs, and the TaN resistor arrays. Btw.: In the schematics, sheet 2/5 of the analogue board, current shunts, there is a strange series arrangement of 'blind' vias. These solder joints probably serve as thermocouple balancing for the current resistors, to achieve low thermal voltages. That's not so easy to be replaced in lead-free technology, either. The only board agilent did redesign, very probably lead-free, and in a compatible / plug-in replacement manner, was the µP board. Up to then, it was easy to buy and insert four 62256 SRAMs and get the 01 option for maybe 30€. I've done that directly in 1989, for my first 3458A @ university, later in 2009 for my own hp3458A (2001 vintage), and these RAMs are still available today, if you have an elder unit. The new µP board has the memory inside the new ASIC, and that has to be enabled by a 8 pin IC, maybe a PAL or such. For the sake of tradition, KS still takes a lot of money for that option, but you might order this PAL for about 30$, if I remember right, and get this option for a reasonable price.. but that's not 'official', and maybe KS meanwhile has shut down this loophole. Please visit TiNs exceptional website (xdevs.com) for further details Frank ___ volt-nuts mailing list -- volt-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts and follow the instructions there.
[volt-nuts] What's the probability of a random used 3458A passing a Keysight calibration?
Surely the AC errors if any must be associated with the A2 AC Convertor board - 03458-66502? >> Dave Why should this be a hardware error??? It's much much more probable, that this is a natural drift phenomenon. This is 'as found' and the 'as left' report is missing, that would tell the whole story. Anyhow, this can very probably be adjusted by the regular calibration process. Frank ___ volt-nuts mailing list -- volt-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts and follow the instructions there.
[volt-nuts] What's the probability of a random used 3458A passing a Keysight calibration?
I also saw this, so at least, they will have adjusted this range. Yes you're right, the ADC is the A3 board. I confused that with the A5 controller board, which is the only one which had been redesigned to SMT components. There's no chance to identify a refurbished or replaced board, other than opening the case and looking inside. And even then, it's not evident, that the ADC IC is ok. You only might do the test according to AN-18. Replacement boards will have the same part number, which changes only by redesign. Frank ___ volt-nuts mailing list -- volt-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts and follow the instructions there.
[volt-nuts] What's the probability of a random used 3458A passing a Keysight calibration?
I meant to say, that the ADC ASIC determines crucially the stability.. and this special serial number US28032500 is eventually affected by the AN-18 service note. So that may be the reason, why they claim 2100$ for repair, i.e. replacement of A5, the ADC PCB.. and that had nothing to do with that ACV outlier. The "as left" data is not displayed, probably the instrument was completely adjusted afterwards. The cal report is also not online, pity. So I wouldn't touch that instrument at all, as the ADC drift problem (CAL?72) is not checked. Frank ___ volt-nuts mailing list -- volt-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts and follow the instructions there.
[volt-nuts] What's the probability of a random used 3458A passing a Keysight calibration?
David, that instrument on ebay is definitely from the agilent area, i.e. built after 2000. It is really ok, and does not need any repair. Obviously, this AC range is a bit outside calibration @ 3V, 4MHz, but that is normal after some years w/o adjustment, and not a defect at all. Probably you know the different kinds of calibration at HPAK, obviously they chose the cal w/o adjustment option. Currently, the KS server is down, otherwise one could download the full calibration document, and check that, also the date of manufacturing . If you'd ever need these high frequency measurements, it's possible to do that by means of an appropriate signal generator and a thermocouple. The internal LTZ1000A circuit might stabilize with age, also the internal 40k VHP100 reference resistor might get better. The annual drift of the LTZ circuit is nominally 8ppm, but that is not fully specified. In theory, that applies only if the instrument is powered on continuosly. If it's switched off most of the time, the LTZ chip should not age at all, but it may show hysteresis (see AN-18, first item) So to my opinion, in this case it makes no difference, if you have an older or a younger instrument. The rest of the circuit is either (short term) stable, or not, also mostly independant from vintage. Due to the regular AutoCal process, this is of no real concern, as all these other drifts will be cancelled by the machine. The only component which determines the short term stability, is the ADC ASIC. It might have a good T.C., from zero to max. 0.5ppm/°C, and it might have a low timely drift in its calibration parameter CAL? 72. If latter is not the case, as described in AN-18, 2nd item, you really have a monetary problem. But in any case, this stability risk is the same for every vintage of instruments. Anyhow, I would prefer a newer unit like this one, as some other components degenerate over time, like the un-obtainium fast comparators. In the end, this instrument is not cheap, especially w/o proper adjustment. 3000$/€ would be a more reasonable price. Frank ___ volt-nuts mailing list -- volt-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts and follow the instructions there.
[volt-nuts] GPIB Issue with 3458A
Randy, I doubt, that your hp unit is defective, as it might work under certain circumstances, and also the FW version 8.2 vs. 9.2 will not make any difference, as only a few pair of bytes are different. All versions of the A5 board are designed to be completely interchangeable over all versions of 3458A instruments. I have another theory, which deals with the hardware version difference between the agilent and the hp unit. Your agilent unit might already have the 66547, full SMD board for A5, whereas the hp will for sure have one of the old 66505/515, through- hole boards, Rev A, B, or C. Illya has all of them pictured in his first 3458A repair blog. All the old versions (A, B, C) have GPIB bus transceivers, TI 75ALS160 for the data bus, and NS 75161 for the handshake signals. The new SMD board has TI 75ALS160 / TI 75ALS161, instead. So, the handshake bus uses a different logic family for the new board, and therefore, the logic levels / immunity might be slightly different. My hp3458A has got the REV C A5 board, and it's from 2000. Since 2009, I run a similar datalogging program for capturing LTZ1000 output versus time, each 4 seconds, on an old PC with WIN98, Turbo Pascal, an XT bus GPIB card from CEC, with the NEC chip. This program and hardware ran properly for years, (decades), until more and more errors during the 24h acquisition runs occured, vaguely reminding me of your trouble. The output values were always correct, but frequently, 10sec time-outs ocurred, giving zero readings in the datalog. This time-out is obviously related to the handshaking mechanism. Finally, I found out, that the old ATX PSU was failing, especially its input snubber network was gone defect, which probably created big EMC spikes. After replacing the PSU, all these errors vanished completely. My conclusion was, that the GPIB bus is susceptible to such EMC disturbances, although it's solid TTL logic. As obviously the handshake signals were affected only, their higher susceptibility to disturbances may be caused by the difference in the logic family of the transceiver IC. The agilent 3458A might be more tolerant to that, having the ALS161 version, instead. So you might also look for EMC problems in your lab, like SMPS, LED lamps, or a defective PC power supply. Maybe, these Prologix adaptors (clones?) have critical logic levels, especially on the handshake bus, or an improper ground. You might search for these signal problems directly on the GPIB cable. Frank ___ volt-nuts mailing list -- volt-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts and follow the instructions there.
[volt-nuts] Unable to zero 3420B on 10**5 sensitivity
David, I've thought a bit further about this excessive offset of at least 8µV. That could be simply due to excessive offset of the DC amplifier part of the meter amplifier, caused by ageing, i.e. drifted resistors or faulty elctrolytics. The chopper has about 5000x ac gain, due to rectification, that's about 2500x effective dc gain. So these 8µV translate into 20mV too much of dc offset of the dc amplifier. How the 1V zero (R16) affects the DC amplifier offset, I don't comprehend yet. But the next step is to check the DC amplifier. Frank ___ volt-nuts mailing list -- volt-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts and follow the instructions there.
[volt-nuts] Unable to zero 3420B on 10**5 sensitivity
David, you should have checked the different parts of the chopped amplifier according to the manual, so to make sure that the DC amplifier works correctly, and that the AC amplifier is also ok, regarding the amplified neon waveforms. But at first you could make another quick check: As you have reworked the zeroing network, make sure by measuring at the wiper of zero pot R1, that you can adjust this voltage between -0.6V and +0.2V. Make sure, that the 1.8M resistor (A1S1R5) has about the correct value. If it driftet upwards excessively, that would reduce the zero trim range greatly. Comparing this schematic with the 845A, which is really very similar, this big un-symmetry of -0.6 vs. +0.2V caught my eye.. in the 845A, it's more like -0.46V vs. +0.56V. As this voltage acts upon the negative input of the chopper (photoelement V2), so indicating that your positive offset is caused by a too negative zero voltage, you could try to short R49, this 47k resistor, to have a symmetric +/- 0.6V. Maybe then you are able to achieve zero at the 10^5 gain. If this is the case, please note the wiper voltage. This is attenuated by 1.8M over 100 Ohm , nominally covering +/- 33µV, which can be checked in the 100µV range. Also note, how the wiper voltage matches with the offset deflection. If that wiper voltage is too high for zeroing, or if it does not match so well with the deflection, there may be something wrong in one of the amplifiers, or there is excessive leakage current, (by contamination by fingers of pre-owner), or else... That is to be analaysed later on. Frank ___ volt-nuts mailing list -- volt-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts and follow the instructions there.
[volt-nuts] HP-3458A NVRAM
Randy, 'quarks' is another volt-nuts companion. His instrument is from 1989, and we both changed all nvSRAMs in our 3458As. The complete designations are: DS1220AD-150+ and DS1230Y-150+. If the 1230 battery fails, you will get failure messages on booting the 3458A, and your stored settings and key function definitions will be lost, so that's a bit uncomfortable at most. Frank ___ volt-nuts mailing list -- volt-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts and follow the instructions there.
[volt-nuts] HP-3458A NVRAM
Hello Randy, I have replaced the CAL-nvRAM by a DS1220AD-150 in 2013, as the Y version was not available. The speed was not the problem, but a slightly different supply volatge range, if I remeber correctly. My 3458A works w/o any problem, as does the instruments of 'quarks', who got the 2nd nvRAM which I have ordered. Frank ___ volt-nuts mailing list -- volt-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts and follow the instructions there.
[volt-nuts] 3420B chopper drive problem
Another probable error in the manual, on page 5-8 (80), diagram, which shows the monitoring of the neon currents: That diagram is ok for the A-version, as both neon currents are measured with the same polarity in its oscillator circuit. The B-version anyhow should look different, as the currents should be of different polarity, so a positive pulse with about +150mV, and a negative pulse with -75mV (or vice versa, depending on the polarity of the diodes), always a signal pause at 0V. One neon current is monitored over 51 Ohm, the other one over two 51Ohm reistors in series, therefore these different amplitudes. Frank ___ volt-nuts mailing list -- volt-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts and follow the instructions there.
[volt-nuts] 3420B chopper drive problem
Hello Dave, in the schematic, the polarity of the diodes in series with the neon bulbs is definitely wrong. If you compare with the 3420A on page 7-7 (29), but also with the Fluke 845 which is similar, both neons have to be switched at 180°. Maybe someone else has erroneously reversed one of the diodes in your unit before? If both neons are really firing in phase, instead, that could explain this non-symmetric waveform, which you observe. Frank ___ volt-nuts mailing list -- volt-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts and follow the instructions there.
[volt-nuts] What is Transfer Accuracy?
I just want to add, that this parameter for me is the most important one, for such instruments like the 3458A, the Fluke 720A and the 752A. Latter ones are explicitely called "Transfer Standards", and therefore you see, that this parameter has to be well defined, and not being "fuzzy". The transfer accuracy is always mixed with the linearity (where I come back to my problem of definition) Such real "Autocal" instruments (plus Fluke 5440, 5720) simply "live" from their own capability of making ultra precise transfers Outstanding are the 3458A and the 5720A, because they not only do transfers within a single mode by their linearity (i.e. auto-calibrating all DCV ranges), but also to different modes (ACV, ACI, DCI) by ultra precise converters (mainly AC/DC). Here you have to specify also the transfer accuracy of the conversion method. Frank ___ volt-nuts mailing list -- volt-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts and follow the instructions there.
Re: [volt-nuts] 3458a RAM batteries - longer life if unit is continually powered?
Hi Poul-Henning, you've misunderstood, what I've done.. I simply searched the text fragments directly in the binary file, with a HEX-editor. I did no SW-reverse-engineering! I just wanted to identify, which failure (text) messages come up, when either the config RAMs, or the cal RAM fail... Nobody up to now reported the correct failure messages in case of failure of the 8kB CAL RAM. But perhaps you can tell from your reverse enginieering, what the error message would be? regards Frank Am 27.03.2014 19:11, schrieb Poul-Henning Kamp: In message <533452cd.10...@freenet.de>, Frank Stellmach writes: So I've scanned the text fragments in the firmware file: I have the firmware reverse engineered, and it is a fair deal more complicated than you posit, there are both checksums and hardware mechanisms involved. ___ volt-nuts mailing list -- volt-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts and follow the instructions there.
