|
Hi brain trust,
As anyone else had a 2712 suddenly go deaf? The input has become extremely insensitive; this has happened temporarily a few times in the past, but cleared up when the unit warmed up for a few minutes...not today. Before I tore into it I thought I would ask around. I enabled CAL DEBUG FLAGS and ran a Normalization and most of the tests failed: ----------------------------------------------------------------- AMPLITUDE NORMALIZATIONS 9 GAIN AND ATTEN FAIL N 1 GAIN CORRECTION DAC FAIL N 2 FULL SCREEN OFFSETS FAIL N 3 BW FLTR AMPLTD(LIN) PASS N 4 BW FLTR AMPLTD(LOG) PASS N 5 PREAMP GAIN FAIL N 6 LOG AMPLIFIER FAIL N 7 SENSITIVITY PASS N 8 VR FILTER BW & CEN FAIL N 9 ALL AMPLITUDE NORMALIZATIONS AMPLITUDE NORMALIZATION FAILED --------------------------------------------------------------- It occurs to me this could most likely be first in the series chain of relays in the attenuator section at the input (K110 or K120). A high signal level could affect the contact plating integrity, causing intermittent contact. The TO-5 relays are Tek p/n 148-0198-00 and apparently unobtanium, but a suitable replacement is the TE TE M39016/15-077P 6V 98 Ohm DPDT Non-Latching Low Signal Relay. It may not be pin-compatible, but at ~$200/ea seems in the right quality echelon and some are available on eBay for much less. If anyone has had a similar failure in a 2712, I appreciate any info! Cheers, Howard Hoyt |
|
Hello Howard and others,
I have several of these analyzers and took them apart. Perhaps I can help. First I would check the attenuator by injecting a known signal of say 0 dBm 100MHz at the input N-connector, which is connected to the input of the attenuator J100. Then measure with another spectrum analyzer or power meter or whatever you have available, the output of the attenuator at J150. This J150 is located on the right front top side. It is connected to the tubular 1.85 GHz lowpass filter. Go through all the attenuator settings. At max, the output should be within 0.5 dB with the said 0 dBm. Check this first and come back to us here. There could go something wrong with the LO level too, like I had. I assume you have he 2711/2712 Sevice Manual from What kind of suitable measurement equipment do you have at hand? BTW. Sorry Edward for me saying, but this 2712 analyzer is quite different than the series you mentioned. Jaap Netherlands |
|
Dear Howard,
Many thanks, I suffered from 2710, horrible interfaces and keypad, easy to blow input. 2712 may be better. The relay may be recovered without opening up. Most small Tek relays,can be "exercised" Remove from PCB Get a,mains powered LV transformers rated for relay coil V, use a diode to make pulses. It should buzz when connected. Take a resistor box or power rhéostat in series with contact and another LV transformer Run 5....50 ma thru contact Idea is to run substantial currents for Burning off oxidation Try it Finally your relay can be found on eBay or forums. Get tek parts list, translations form Tek PN to mfg eg teledyne Bon chance Jon |
|
Jean-Paul,
Ah yes I have done that to relays in the past as well, the buzz cleaning...I thought you meant to open up the relay...although I have to admit seeing the construction of such a tiny DPDT relay under a microscope would be very fun! Do you have a link to a table showing Tek P/N to Industry P/N equivalent? Jaap, Checking signal level through the attenuator will certainly be my first step, thanks! I have an Agilent 33120A sig gen and some HP 400EL meters which are pretty good up to 10 MHz or so, above that I use my Tek 2465... First i have to get the repairs done which occupy the bench, then it's time to pull the 2712 open. Thanks for all the suggestions, I will report back what I find! Howie |
|
On 2024-04-17 05:36, Howard Hoyt wrote:
Drawings of inside: Arie |
|
Hi folks,
Upon extracting the RF Attenuator assy from the 2712, I see the TO-5 cans themselves are soldered to the PCB, as is a shield over the bottom: /g/TekScopes/album?id=294504 Since there is a preamplifier IC in line I want to try merely exercising the relays using a 5 V 10 Hz square wave first without the 10 mA current source to sputter oxides. If this doesn't work, then I'll proceed to hot-air unsolder the cover and relays and do a proper contact sputtering using the 10 Hz buzz with 10 mA current source... I'll report back. Howie |
|
Hi Jaap,
The calibrator signal was getting through correctly, but RF from the front panel jack was not, so I suspected K110, the first relay in the attenuator chain which selects the front panel or calibrator signal. I spent the better part of an hour unsoldering the shield can from the PCB...and the technique I found that worked the best was to take a small Metcal chisel laying along the junction between the two melting the solder and with the other hand keep a Hakko 808 vacuum sucking at the junction. It really cleaned the solder out well, but it was still a hassle breaking the remaining solder joint and removing the cover. Upon disassembly (picture here: /g/TekScopes/photo/294504/3774186) K110 ohms out correctly, of course I had buzzed the coil beforehand, so that may have (temporarily?) cured it if that was indeed the problem. That will teach me to break a troubleshooting rule: change one variable at a time. I plan to temporarily put the attenuator back in line and retest before soldering the cover back on. Thank you for your assistance! Howie |
|
Hi Jaap,
I'm back on the 2712 diagnosis... I tested all the relays in the RF Attenuator block...and they were good, but upon reassembly the issue still exists: it does indeed show input, but the level is some 24 dB down, frequency dependent. I had thought the Calibrator signal was proper, but the unit fails just about all the Amplitude Normalizations, so the problem likely exists with the calibration signal as well. Since the RF Attenuator relays are OK, the issue must be afterwards, in the 1st Converter, LO or some other block I do not have test equipment to check, I only have a 350 MHz scope... Barring any further ideas from the group, is there anyone in the US Southeast who repairs these older Tek gear? Thanks for the help, Howard Hoyt |
|
Hi Jon,
The 2712 has been in dedicated power supply ripple QC duty. It is fed by a jig with 0.1 uF caps blocking the 14 V supply DC, so it only sees any RF on the SMPS supply rails. It is possible that one of the supplies being checked had a bad output filter and very high ripple, but I do not remember one. Well, if that first mixer is burnt out I wonder if the CR256B or CA130 diodes have failed? I did a search through the archive for 2712 issues and did not find any other reports where the fault was found. I guess I'll have to pull the can and do some measuring. Cheers! Howie |
|
Yes, the 2712 is different from the 492 series. But it still has a 1st if via a YIG but now at a slightly higher 2110MHz, 100MHz master and the same 110 2nd if etc. It is a completely new physical design, newer CPU, maybe about 2-3 times faster (my estimate) than the 68xx in the 492. But, the general RF design is similary, phase gate, inner, outer loop control, analog log amps, 2 frequency markers and all that.
The relay atten is a much less expensive design than the uwave cavity in the 492's which works up to10 times the frequency using thin film wafers. (The 492 attens are not repairable but can be found on paybay). As a side note, the 275x 279x series parts are way way more expensive than 492 parts (ignoring the 40GHz model) I wonder what 2712 parts costs are like? |
|
A cheap $100 vna can help diagnose signal path issues. This shows what happens when a micro crack in a connector can do.
/g/TekScopes/photo/284401/3583773 |
|
Hello Howard and all,
As Jon mentioned it can very well be that the mixer diodes have been destroyed. Below are some level problems I encountered in some of these analyzers. Please check Q350 - 2N3904 (SMD-marking 1A) in the VR Module. I had a broken one. This transistor gets too hot under normal circumstances because of the combination of UCE en IC. The UCE is about 11,5V and the IC is between 20 and 25mA, in my case 23mA. The dissipation then is between 250 - 300mW and that is a bit too high for a SOT-23. In my case I put in a new one and I cooled it via a heat pad and a small 1”x 1” aluminum heat-sink of the right (I forgot) thickness against the inside of the lid of the VR Module. Another thing can be a too low level LO signal coming from J380 from the 1st LO Buffer Amp. The level should be about +13dBm, but in one of my analyzers this was way lower. Please measure that if you can. The problem in my case turned out to be caused by two leaking SMD electrolytic capacitors inside this hard to reach unit. Residue from the leaking (a bit of black sticky fluid) was hiding under a SMD capacitor C292. Under this resistor there was a high impedance track going to the input of the opamp that stabilizes the LO output level a bit to a certain level. In the end I removed all the following components that I had my doubts on in that unit: SMD components C292, C298, C828, R280, R166, R168, R266, C267, C160 and R264. I cleaned everything and put them back. The problematic SMD electrolytics C180 10μF/50V and C392 10μF/50V were replaced by new ones of course. Good luck and please keep us updated. Jaap |
|
Thank you Jaap and Jon,
Excellent troubleshooting advice! Looking at suspect parts the one I think maybe a bit difficult to source is the CR130, which is a Tek manufactured Schottky diode quad p/n 152-1063-00. I looked through the Tek Semi cross reference: tek_xref_free.pdf, but I think that doc is pre-1990 so no diodes above 152-0800-00 were listed. Anyone have a newer Tek semi cross reference/databook? If it does turn out to be a failed part I may try some 2 ns SOT-23 Schottky diodes in its place. I will report back after the next bit of troubleshooting. Cheers, Howard Hoyt |
|
Hi Japp, all,
After borrowing another broadcaster's SA to knock out a bunch of work I am back on my 2712. I fully tested the Attenuator, taking the output to a scope and exercising the Ref Level settings and it was working properly. I checked Q350 in the VR Module, and it was good, which leads me to suspect the RF Input board or 1st Converter, the module which is so tough to get into...30 screws.... I wish I could check the LO level into this module, and I'd use the 2712 SA, but...the highest frequency I can check otherwise is 350 MHz with a Tek 2465, or ~700 MHz with the RF probe on my HP 410C. Thanks for all your help folks, this is not an easy problem without better instruments... Cheers, Howard Hoyt |
Jean-Paul
Arie de Muijnck