For sentimental reasons I've been trying to repair my old TLS 216. I know this doesn't make sense but it's been the first scope I bought myself (not knowing what a mess the 75R inputs and the probes are). But I got several probes with it and for years this was my fastest scope - not that I would have needed that 500MHz.
My original ACQ board was so corroded, I couldn't get it running even with fixing more than 20 corroded traces and checking everything over and over again. Well it was sort of running except for the second DAC with channels 5..8 but then I shorted some DAC outputs to VCC or GND when probing around, now it's silent forever. You shouldn't probe around when you're in a hurry.
Occasionally I got another scope for parts with a defective display and the problem of crumbling isolation of the input ports. That ACQ board was in much better condition. Again all caps have leaked but the corrosion was not heavy, no pads falling off or corroded/broken traces. Two areas next to the 10uF caps to clean up a little but nothing special. Interestingly the caps on my old CPU board haven't leaked while on the new one they have. Good thing: the TLS comes up and after SPC the offsets are gone but it doesn't pass self test if all DIPs are off.
I think the 2nd TLS wasn't really running correctly (though it was claimed to be) as it was coming with DIP SW 3 set to on - which skips most of the self tests. However I haven't checked that before re-capping. I assume that has only been done as there has been an issue or the CAL NVRAM didn't match the ACQ board.
I always wondered why all these CPU boards come with the same EPROM (both TDS and TLS have 160-9335-00) and now I think I found that the firmware is mostly contained in the NVRAM, together with the CAL data. For my mainboard both the RTC and NVRAM are overlapping at 0x0400.0000. RTC on the first 16 bytes, then NVRAM. You can see this in the schematics for the TDS 544A from
http://hakanh.com/dl/kits.htm on A11.p4. NVRAM is selected with U1055 if A26:A24 == 100b. R102 is not populated so the OR of A19:A4 selects NVRAM2 (the 512kB NVRAM) if any of these bits are 1 and NVRAM1 (the RTC) if all bits A19:A4 are zero (first 16 addrs). So with a 512kB NVRAM dump one can save software, CAL data, options, logs, everything.
It seems that different versions of the 671-2413-xx mainboard (even with completely different layout) behave the same, at least my two boards are compatible though one has the Dallas chips on top and the other one on the side. However there might be different versions, I've seen boards on ebay where R102 is populated and U170 is not while on my board it's the other way. But my two boards 671-2413-00 and 671-2413-03 are compatible.
The original mainboard contained software 1.0e and with this software/CAL data, both sampler vertical and sampler horizontal failed in SPC or after POST.
The other mainboard contained software 1.0.3e and with this software/CAL data only sampler horizontal fails.
Unfortunately the ACQ boards are slightly different, the old one was 672-2552-00 and the other 672-2552-02. Though I don't really see a difference there. I copied the NVRAM (software + CAL) to the other board and now I've got 1.0.3e on both CPU boards (floppy method as described below).
The scope basically works on all channels but it has increased jitter and the trigger is approx. 5ns late, both in first and second time base. I checked the time interpolator/jitter ramp modules, the little sub-board on J401 (1st time base) and J402 (2nd time base). The module seems to contain the function of A10.p13 of the TDS544A except for U1601A. The buffered RAMP comes from the ACQ board. The pin assignment is
1 ... +15V
2 ... -15V
3 ... -5V
4 ... /RAMPST in -> VR113 -> Q116
5 ... RAMPST in -> VR115 -> Q117
6 ... GND
7 ... RAMPSW in -> Q101
8 ... /RAMPSW in -> Q102
9 ... +5V
10 ... LM311.7 out = EQRAMP out
11 ... GND
12 ... RAMP1 buffered in ... should be -8V .. -1.8V but is 0V on my scope (for both 1st and second time base)
So even the part numbering matches if you subtract 1500 from the TDS544A schematics (Q1603 -> Q103). The issue that the same 5ns show on both timebases (checked with first on rising edge, second on falling edge) indicates that this isn't an issue of the dedicated time interpolators.
RAMP1 and RAMP2 come from a "daculator" (IMP I10412-01), basically a serial SDI/SDO 16 channel DAC as shown in the TDS520 service manual with schematics (document 070971003 from Tektronix) on page 275. I checked that daculator, it's U600 next to U620). Serial data and clock is present, +/-5V supply and -2.5Vref are OK - though I didn't find where the -2.5Vref comes from. Supposed to be the TL074 next to it isn't. Maybe the use an internal reference. All outputs show some kind of signal but pins 28 and 1 (out 13 and 14) that are the RAMP input for the time interpolators are on 0V. Though they are not shorted to GND and show some voltage during POST of the device. So they are turned off by the software, probably as SPC or POST fails. I didn't find such parts in the TDS544A, these scopes seem to use the typical Tektronix S&H with lots of 4051 CMOS multiplexers attached to one AD667 DAC. The TLS uses 9 of these IMP I10412 instead (one common and 4x2 for the ADC groups).
The PLL around the 40MHz oscillator and MC1204DFN U504 seems to be OK (that's even the same part number in the TDS 544A).
So my questions would be:
- Does somebody have a hint what that "ERROR: 2260 Calibration failed Sampler Horizontal" could be? I assume there should be some kind of sampling of the horizontal time base but I can't find how this is done. I'd expect some digital operation in the two ASICs U620 and U650. Unfortunately the TLS is so different from the TDS in the control part as it doesn't use that 68HC05 controller but maybe that two special ASICs. Currently I don't have an idea where a problem could be. On the TDS544A the EORAMP or EQRAMP signals run into the ADG225C (p21 EORM and p24 EORD). Here it's the same, the signals are routed to the first ADG225 U350. The inputs on the second ADG255 U660 are shorted to GND. So maybe the sampling happens there.
- As the "Sampler Vertical" error vanished with the firmware upgrade or CAL data replacement, maybe the issue could be solved with a different software version or NVRAM dump. If some owner of a TLS 216 could share his NVRAM data this would be highly appreciated. I didn't have good results trying to read the NVRAM with a programmer. The GQ-4x4 failed completely. The TL866 seems to start too fast after power on so the first 100..1000 bytes are lost. And it read 4 times the same content which cannot be the case if the floppy NVRAM dump is correct.
- Does somebody know if there is a similar calibration software for the TLS216 as for the TDS series? I couldn't find it. Or some hints which NVRAM areas to touch or manipulate for the relevant CAL constants?
The method described in
https://www.eevblog.com/forum/testgear/tektronix-tds500600700-nvram-floppy-dump-tool/ worked perfectly and does not require the NVRAM to be removed. Not the tdsNvramFloppyDumper, that application cannot be launched and is ignored. But tdsNvramMinimalFloppyDumper works like a charm. Even the write back works. I think this is a must have for all TLS/TDS users with the floppy option. Maybe a good idea to create a backup of your NVRAM data now. Even though I'm surprised that the Dallas chips still hold date, time and data after more than 25 years (boards are from 1993). So if somebody with a TLS216 and maybe an 672-2552-00 ACQ board can share his CAL data, I could try this. Maybe it's just an issue with some calibration constants and no hardware issue and the device could pass with the correct configuration. At least I'd give it a try.
The 5ns latency and the 500ps jitter don't really matter as I rarely use the device and if then only for low speed tasks. So basically I consider the repair as half success. But on the other hand if it could be fixed completely, this would even be better.
Home
Help
Search
Login
Register
