TWD120 up and running!
After several false starts, hiccups and a gaffe, I have the TWD120 graciously provided by Charlie Conger fully operational.toggle quoted messageShow quoted text
I won't go into great detail but will touch on all the issues I dealt with.
Getting the TWD120 to pass self test was iffy. I found that after letting it run overnight, probably refreshing the more than 20 years old filter caps somewhat, it began to pass consistently. Before it started to pass self test, I opened it up and poked around a bit with my most handy analog 'scope (A T922 I had previously restored). I noted what I thought was more noise on the supply rails than I thought was ideal, but, considering I had no idea what I was looking for, all looked reasonable.
I initially attempted to use Windows 98SE for an O/S but I could not get the drivers for the Adaptec SCSI card to load. Boo. I was able to get drivers for Future Domain and Symbios cards to load but had SCSI communications problems between those adapters and the TWD120. Tektronix does specifically call for an Adaptec card.
I installed Windows 2000 on the same machine and had no issues getting the Adaptec drivers working. Tried to load version 3.1 of the TWD120 software that Charlie included with it. Got more SCSI communications errors. In sort of a last try, I installed the 1.01 (initial release) software that I downloaded from the Tek Wiki. That worked! Hooray. Now it even acted like an oscilloscope.
However (Seems there is ALWAYS a "however"), there was too much noise in the trace. It didn't change amplitude as I adjusted input sensitivity. Hmmm. I sped up the sweep rate and realized I was looking at residual signal from the switching supply. Opened it up again and found a nominally +20vdc rail with more than 50% noise. Pulled the supply can lid off and removed the whole PCB from the chassis and poked around the supply a bit. I traced the noise back to a 10uF electrolytic fed directly from the pulse transformer via a Schottky rectifier. Pulled that cap and tested it: Yup, it was open. Replaced it. Got to thinking that there was also some noise on the other rails, although much decreased. There are 17 (seventeen!) 470uF/35vdc caps inside the supply can. I pulled as few and they all read almost exactly 400uF, which at first glance would not seem to be terribly bad. However, my experience is that good capacitors will tend to reform up to a higher value if they live at a substantially lower voltage for very long (approx: +-5v, +-9v, +20v and +16v input). I would normally expect a greater capacitance rather than reduced. As an educated shot, with windage, I ordered new low ESR caps from Mouser. Since they would fit, I went with 680uF instead of the original value.
After replacing al 17 of the 470uF caps, I restarted the TWD120 and found all supply rails to be absolutely devoid of noise and a tiny bit higher than before. Self test passing took about 1/3 less time as before.
Then, I did it..... I slipped with a probe and shorted the gate of the switching FET: Instant dead. I discovered that Tek engineering, in predictable fashion, used a cool new style power FET called SensaFET that was developed by Motorola. Only one problem: SensaFET technology was never a widespread success and was abandoned after only a shot time in the sun. SenseFETs have 5 legs instead of 2. The curious can look them up for details. I started to order the exact part number, but made by ON, but the picture only showed a 3 legged device. I contacted the seller and was assured that this was the correct part number and it did only have 3 legs. Apparently On has recycled at least some old Motorola part numbers. Bad idea! I found a distributor who listed the Motorola version of the part, part of their old stock, who confirmed that it did have 5 legs. I bought two, installed one, crossed my fingers and applied power to the TWD120. Up came all the rails and self test was passed. It only cost me an extra $15.00@ for rare $0.75 parts, research time and a week+ delay to recover from my gaffe. (Sheesh!)
Even though it's an 8 bit system, the HiRes mode provides a clean enough display that I can use the TWD120 in my audio work.
I would like to know what sort of improvements Tektronix came up with in getting versions of the software up to 3.0 out the door. If there are any current or recent Tek employees out there, perhaps they could inquire about the availability of any or all of these upgrades.
I did an ink over pencil hand draw of a VERY partial schematic of the TWD120 power supply. I can send a scan of that to any list member who requests it.
Aural Technology, Ashland, OR
By my calculation, the dynamic range of the universe is roughly 679dB,
which is approximately 225 bits, collected at a rate 1.714287514x10^23 sps.
On 11/3/20 4:49 PM, Siggi wrote: