comparison: content/hp8904a.rst
content/hp8904a.rst
- changeset 24
- 89689c6e9cf4
- child 44
- 96b26fb52f0b
equal
deleted
inserted
replaced
| 2:9d2c5ded6441 | 24:89689c6e9cf4 |
|---|---|
| 1 =============================================== | |
| 2 HP8904A Multifunction Synthetizer Restoration | |
| 3 =============================================== | |
| 4 | |
| 5 :Author: David Douard | |
| 6 :Date: 2016-02-20 | |
| 7 :Category: Electronics | |
| 8 :Tags: HP8904A, repair, test equipment, signal generator | |
| 9 | |
| 10 | |
| 11 What Is This About | |
| 12 ================== | |
| 13 | |
| 14 I have a couple of HP8904A multifunction synthetizers for years | |
| 15 (bought them on ebay around 2007 IIRC). One of them is working | |
| 16 properly, but the second one had very inaccurate output levels, | |
| 17 generating a non-symmetric signal when output level was set above 5V | |
| 18 or something like that. | |
| 19 | |
| 20 .. image:: {filename}images/hp8904a/twin_HP8904A.jpg | |
| 21 :alt: My 2 HP8904A synthetizers. | |
| 22 | |
| 23 I knew for a fact that the PSU was the problem (thanks to the fact I | |
| 24 have 2 units, I could easily swap elements between the units). Back | |
| 25 then, I spent a bit of time trying to figure out what was going wrong, | |
| 26 and detected the +16.3V output of the PSU was failing at around 9 or | |
| 27 10V. I did not had a chance to investigate more then. | |
| 28 | |
| 29 For some strange reasons, I've recently decided to fix this PSU | |
| 30 problem. | |
| 31 | |
| 32 | |
| 33 The PSU Problem | |
| 34 =============== | |
| 35 | |
| 36 .. image:: {filename}images/hp8904a/hp8904a_top.jpg | |
| 37 :alt: Top view of the HP8904A | |
| 38 | |
| 39 The PSU board is the one at the rear of the unit, on the upper part of | |
| 40 the case (the lower part hold the analogic boards). The PSU is not a | |
| 41 custom HP board, but is a Computer Products switchmode supply (90 | |
| 42 Watts, Model XL51-5601). | |
| 43 | |
| 44 It delivers 4 voltages on 2 separated ground planes: | |
| 45 | |
| 46 - a +16.3V and -16.3V rail (with their own ground plane), | |
| 47 - a 5.1V for logic devices, | |
| 48 - a -15V (same ground as the +5.1V) which powers at least the fan. | |
| 49 | |
| 50 There are not too much material on the web about the possible PSU | |
| 51 problems of this equipment. A few threads on forums and | |
| 52 `this very interesting blog by Simon Schrödle`_ | |
| 53 | |
| 54 .. _`this very interesting blog by Simon Schrödle`: http://www.simonsdialogs.com/category/repairs/hpak-8904a/ | |
| 55 | |
| 56 But his problem was different from mine. | |
| 57 | |
| 58 On my device, the problem was with the +16.3V output falling at around | |
| 59 9V *on load*. The level was at his proper 16.3V level when the main | |
| 60 board as not plugged in. | |
| 61 | |
| 62 I first gave a close look at the caps, looking for some evidence of a | |
| 63 failing capacitor... which is always the culprit to think of in this | |
| 64 kind of situation. But I could not find anything obvious. | |
| 65 | |
| 66 So I started to reverse engineer the circuit to try to understand a | |
| 67 bit the possible causes. | |
| 68 | |
| 69 The PSU consists in 2 PCBs, the main one which is a double-layer PCB, | |
| 70 and a small daughter board with several op amps and zener | |
| 71 regulators. Obviously the regulation stuff. | |
| 72 | |
| 73 In order to be able to follow the tracks, I had to remove the heat | |
| 74 sink and split these two boards apart, since it's not easy to deal | |
| 75 with this: | |
| 76 | |
| 77 .. image:: {filename}images/hp8904a/hp8904_psu_followme.jpg | |
| 78 :alt: part of the +16.3V PSU rail | |
| 79 | |
| 80 So I removed the heatsink and disassembled this small board from the | |
| 81 main one (which was quite painful, since I don't have a disoldering | |
| 82 station. Disassembling the 3x 8 pins angle connectors from the double | |
| 83 sided PCB was not a piece of cake...) | |
| 84 | |
| 85 .. image:: {filename}images/hp8904a/HP8904A_daughter_board.jpg | |
| 86 :alt: the small daughter board of the PSU | |
| 87 | |
| 88 After that, I noticed a possible leak of a small cap, but this was on | |
| 89 the 5.1V rail. Should not be related to my +16.3V problem. | |
| 90 | |
| 91 But I decided to take advantage of having disassembled the PSU to | |
| 92 replace a few of these caps, starting with this dubious one. | |
| 93 | |
| 94 I also decided to replace the soldered assembly for the two PCBs by | |
| 95 some connectors so I can easily install and remove the daughter board. | |
| 96 | |
| 97 After having replaced the caps, the PSU was still failing. So I really | |
| 98 had to understand the schematic to investigate more. | |
| 99 | |
| 100 The part of the schematic involved in the +16.3V rail is the following: | |
| 101 | |
| 102 .. image:: {filename}images/hp8904a/psu_schematic.png | |
| 103 :alt: part of the schemtic of the PSU | |
| 104 | |
| 105 At first glance (the schematic was then a bunch of poorly made | |
| 106 "Dave-CAD-like" sketches on several sheets), the exact role of the | |
| 107 LM392 opamp was not clear to me, but when I redrew the schematic a bit | |
| 108 more clearly, it became obvious it was some sort of current limitation | |
| 109 circuit. And, sure, my problem could very well be a current limitation | |
| 110 system being a bit zealous. | |
| 111 | |
| 112 So I started to probe the resistors around the LM392 to check their | |
| 113 values. And guess what? R7 was dead. It's a 20k 1% resistor, and it | |
| 114 was not letting any current pass through. | |
| 115 | |
| 116 What are the odd of such a failure on a PSU like that? Unbelievable! I | |
| 117 was expecting the failure of a cap, of course, maybe an opamp or a | |
| 118 diode, but not a resistor. | |
| 119 | |
| 120 I have not yet bought a proper 20k stable 1% resistor (which I don't | |
| 121 have laying around), but I've installed a pair of 1% resistors in parallel | |
| 122 (can't remember the exact values I've used) to get a close 20k. And bingo! | |
| 123 the HP8904A can output a nice 10Vrms again! | |
| 124 | |
| 125 Of course, while I was digging around my PSU problem, the Schaffner | |
| 126 power line filter died in a violent smelly smoky way (took a while to | |
| 127 get rid of the smell). | |
| 128 | |
| 129 Battery, RAM and software options | |
| 130 ================================= | |
| 131 | |
| 132 One of my goals when I looked into my two HP8904A was to check the on | |
| 133 board batteries responsible for keeping the content of the memory. | |
| 134 | |
| 135 This is a critical thing, since the content of the RAM must NOT be | |
| 136 lost otherwise the device won't work anymore (the official procedure | |
| 137 to change the battery involeve sodering a power source in parallel to | |
| 138 ensure the RAM chip is always under power). | |
| 139 | |
| 140 The HP8904A seems to be the first HP test equipment to have | |
| 141 software-enabled options. The option activation depends on the serial | |
| 142 number of the machine, which is also stored in this "non-volatile" | |
| 143 RAM. | |
| 144 | |
| 145 There is mainly one option one does not want to loose, it's the | |
| 146 Option 1. It enables the Channel Configuration Mode Enhancement. It's | |
| 147 the one that make the unit so handy (allowing to mix up to 4 | |
| 148 waveforms). | |
| 149 | |
| 150 I already cheated a bit back then. In fact, one of my 2 HP8904 came | |
| 151 with no option, the other with option 1 and 2. Option 2 means it have | |
| 152 a second analogic board allowing two separated outputs. So I used the | |
| 153 serial number of the unit coming with option 1 on the other one, and | |
| 154 bingo, this latest had then the option 1 activated also. | |
| 155 | |
| 156 But I recently found on the EEVblog a | |
| 157 `thread pointing to an HP service document`_ describing how to modify | |
| 158 the correct byte in the RAM to activate all the options! | |
| 159 | |
| 160 And the magic trick is:: | |
| 161 | |
| 162 Power-Cycle/Power-On | |
| 163 SHIFT, SERVICE, SHIFT, f4, 0, 9, 2, 4, 4, 8, SHIFT, SERVICE | |
| 164 NEXT, NEXT, NEXT, f1, 3, F, F, C | |
| 165 f2, F, F, f4, f4 | |
| 166 Power-Cycle | |
| 167 | |
| 168 That is: write the value 0xFF at address 0x3FFC. | |
| 169 | |
| 170 The procedure to restore the serial number can be found in the service | |
| 171 manual of the HP8904. | |
| 172 | |
| 173 And in case of failed battery, the memory can be restored following | |
| 174 `this document <{filename}pdfs/HP_8904A_Service_Notes_2.pdf>`_ | |
| 175 | |
| 176 .. _`thread pointing to an HP service document`: http://www.eevblog.com/forum/testgear/hp-8904a-options/ | |
| 177 | |
| 178 | |
| 179 Next steps | |
| 180 ========== | |
| 181 | |
| 182 The first thing I have to do is to replace my temporary fix with a new | |
| 183 stable resistor. | |
| 184 | |
| 185 I'll probably replace the power line filter of the other signal | |
| 186 generator before it dies in a smoky way. | |
| 187 | |
| 188 Then, I need to recalibrate these puppies, but I need to build a bit | |
| 189 of confidence in my HP3456A DVM for this. But that's an other story. | |
| 190 | |
| 191 I also want to replace the fans with some more recent and quiet models. |
