• Please review our updated Terms and Rules here

8272 FDC to 8" drive connections

That's quite an undershoot, Mike! I find it surprising because the READ DATA signal is nowhere close to DSx on the SA800 interface.

I guess the first thing I'd determine is if there's any coupling via the power supply rails. Put a scope on the +Vcc pin of the 7438s and close-by ICs and see if there's the same noise on that. For yucks, do the same with the ground pin on each. Given that you're using 5 feet of ribbon cable with every other conductor grounded, I wouldn't expect significant crosstalk in the cable, particularly not with 150 ohm termination (in comparison, a standard PC-type 3.5" drive uses about 2.2K or more) and only a single line changing state.

I'm pretty certain that this will be easy to resolve. What construction method are you using (i.e. PCB, wire-wrap, PTP, breadboard, etc.)?
 
That's an idea, maybe after the football game. I had also thought of putting an RC circuit on the line maybe 1k and 100pf. The RC time constant of that would allow a quick pulse through, yet suck up the negative spike and maybe a couple of schottky diodes. Originally I had about 11 feet of ribbon and have cut that back to less than half. Every thing I have is wire wrapped. And I have a small ceramic decouple capacitor next to each chip. I'll look for the noise some more. Let you know later this after noon. Thanks, Mike.
 
I'm assuming that your power rails are also substantial; i.e., not simple 30 AWG lines. Where I don't have a ground or power planes, I construct an artificial one with 20AWG bare wire connecting single wirewrap pins, bridged with 104 caps and take my IC power from those. It seems to work pretty well. The power spikes kicked up by some of these old buses can be substantial.
 
My FDC board is built on a Vector S-100 board that has a power bus clad on it. Both the ground and +5 run under each chip and each chip is wired directly to it with short wires. The board has 30, 0.01uf ceramic De-coupling capacitors, mounted near each chip. My regulator is a 7805 with a 0.47uf on the input and 0.01 uf on the output. Seems I'm always battling power problems.

I checked as you suggested and the spikes are on the power bus. Worse yet it's coming from main power. My switching regulator was causing the noise and the spikes. Lots of spikes and damped oscillations. Then as I was trouble shooting, the entire thing failed. Seems the filter capacitor in switching regulator was causing the trouble and then gave up the ghost. Problem is, it also somehow affected my +16 volt supply also. And, I'm wondering if other things were affected.

Well...... now I have something to fix. Two steps back, but maybe the spikes will be gone. Odd thing, just last year I replaced all the big capacitors, for fear they would go bad and then one of the new ones failed, go figure.

Let you know, hopefully not to much other stuff released their quota of smoke. Thanks Mike.
 
I think you're underdoing (is that a word?) the bypass. With good old TTL, I've always used 104s (0.1 uF)--and the 7805 setup seems to be more stable and less prone to spikes, when done like this:

Wph1i.gif


For chip bypass, I generally use MLCC 104 X7R 16V axial-lead caps. They're small enough that they can be slipped on the underside of a socket (there are/were) sockets with them incorporated.
 
Last edited:
Seems that the by pass stuff is not the major problem. Seems my switching power supply that pre-regulates a raw 16 volts down to 8 volts is the problem. When there is no load on it holds 8 volts real nice and has a clean output, but as soon as some load is applied, more than a 1/2 ampere, there are spikes all over the place. A while back I replaced a simple dropping resistor with this pre-regulator. I don't recall whether or not I looked at the loaded wave form or not. The pre-regulator is based on a 78S40 chip as a basic step down switching regulator. I suppose that I'll look at it for a while and attempt to make it work, but I may have to use a different design. One good thing is that all my boards escaped damage. Mike
 
I'm a big fan of the TI (used to be NSC) "simple switcher" ICs. For your application, an LM2576-5 would be perfect; 3A, TO-220 package; very few external components. If you can adapt to a TO-263 package, the LM22670 would be perfect. If you heat-sunk a 7805 sufficiently, you could probably get by with one of those, although more power would be going up in heat than that used by the board.
 
I just tried to post a long text and apparently something timed out. I got kicked out and all the work I did was deleted. I tried to wrote a private message to the site administrator and it happened here too. What am I doing wrong? I'll try again. Mike
 
I'm sure that those components are good and will do the job, but I'm not going to give up on what I have just yet. I have a Fiarchild 78S40 driving a 2N4398 transistor.
SwitchingCircuit.JPG
I believe that I have an idea of what is causing all the spikes on my output. I place my scope across the Collector Emitter of the transistor. I see that the oscilator is running at about 22 Kc and the On time is about 12 uSec and the OFF time is a about 32 uSec and that the transistor is turning on and off very fast. This all looks good.
(I tried to place a picture here) but it refuses to work, I try and place the picture in the next post)
The problem appears to be that when the transistor is ON, it doesn't stay on. At first it saturates at about 1 volt, but rises to 10 volts after the 12 uSec. Something is causing the transistor to slowly turn off. I think that maybe the Inductor/Capcitor/Diode LOOP may be affecting the transistor. I'll be looking some more today, any ideas? Thanks Mike.
 
Just had another idea, maybe I chose the output capacitor incorrectly. When the transistor turns off, the inductor will immediately change polarity to keep the current flowing. The inductor voltage may over shoot and try to ram some extra charge into the output capacitor. If this capacitor has a large ESR this could cause a voltage spike. I'll look for some data on my capacitor. Mike
 
Well, I don't think that the ESR is the problem. Although my Mallory CGO342m055r2c3ph is obsolete, it is advertised to have very low ESR, 12 milli ohms for the one I have. Mike
 
Mike, have you read the app. note for the 78S40?

In particular the appendices are worth looking at. The transistor used in the app note is a 2N6051, which is a PNP Darlington (very high h[sub]fe[/sub]). I don't know if it matters in your application.
 
The application note I've been using is from 1990,. Same an-711 number, but older. I see that this one is 10 years newer. I'll have to read it and see what there difference is. My app note doesn't state a transistor to use, it was my choice. I can see that a greater hfe may have some influence on staying in saturation. It would not be difficult to make a change. Just would have to find one. Still looking. Mike
 
I checked the current waveform through my inductor and it looks pretty good. At least what I had expected.
InductorCurrent.jpg
I looked at the output voltage right at the output capacitor terminals and there is no spikes at all, but as you move farther away from the capacitor the noise increases, especially when you move the ground connection. I'm going to look at re routing the grounding and maybe try a single point ground, if possible.
I also looked at the diode current waveform and didn't see what I had expected. There is noise on it, so I'm suspect of the diode. I had thought that I purchase a couple of these RURP1560 diodes, but I can't find the spares. So I ordered a couple. I also ordered a 10 ampere choke. I figured that I'd make a low pass filter to try and condition the output of the switching regulator. Maybe I can filter them out. So, more waiting time for new parts. Maybe the grounding will help. Mike
 
This is amazing! Sort of akin to alchemy. The routing of the ground wires are very important. I shortened all the leads as much as possible and tried a bunch of different variations of ground connections. Makes a BIG difference. I also did some experiments with a low pass filter. On the output of the switching regulator I connected the secondary of a power transformer and a 3200 uf capacitor. This reduced the ripple and spikes to under 150 milli-volts. I tried a ferrite core torrid, with the same results. I even tried wrapping a length of 20 awg wire around a 3/4" bolt, even this made a big difference. So apparently the amount of inductance is not real important, but there has to be some. Short out the inductor and the noise passes right through. The torrid is 356 uH. So maybe this is the solution to the voltage spikes. I'll have to wait for the parts I ordered and install them, before I can test the FDD line for spikes, but hope is in the air. Mike
 
I have enough challenges. Remember this is tough stuff to an old relay contact logic guy, where the power supplies were 125, 250 and 600 volts DC. Thanks for the help, Mike
 
Creating a PCB board is intriguing. I long time ago, I used to make PCB boards (not many of them) on copper clad board for discrete transistors, resistors, etc. I would use a twice size mask and then reduce it photographically. I would then sensitize a copper clad board with some chemical and expose the board with ultra violet light. This would harden the copper that I didn't want removed. Although this worked, making a S-100 board would be very difficult with this process. I have KiCad and I see that it has some kind of board making program. I hadn't really looked at it much. I'm sure that the program is good, the schematic part is great, but how do you move from a computer output to a completed board? Do you still have to make a mask and acid dip the board? I could try to make a board that I'm sure would not be changed again. A PCB board may be more reliable, but it really restricts any changes that many be needed.

Other items that I learned during this power lesson. This morning I turned on my scope. The probe was still connected to the power bus of the computer, but the computers power cord was not plugged in. Less than 3 feet over head I have an 8 foot florescent lamp. The scope was picking up noise from the lamps. Quite a bit of noise. My 8080A is in a 19" rack mount enclosure, all metal. As soon as I plugged in the cord, the case was grounded and all this noise disappeared. Seems that noise can come from many places. Also, while rewiring some of the power bus, I noticed that some of the crimp on terminals I used were not tight. I used #14 stranded wire for all the connections between the power transformer and the card cage. So, I removed all the wires and checked all the connections, plus I soldered each one. I loose connection can also cause problems. Thanks Mike.

It happened again, in the time it took me to type this note, I was kicked out, and the note was lost. why doesn't this site kick you out so fast?
 
FWIW, if I'm using crimp terminals for power, I crimp, but also solder if possible. Belt-and-braces maybe, but they don't fail.

I've got lots of wire-wrapped prototypes and they work well decades after they were built. When I'm not quite sure if a design is going to work (too many unknowns), wire-wrapping helps me flesh out and tune the design, particularly if it's a one-off board. I can go from pencil-and-paper to the finished prototype in the same day. That's not the case with PCB--doing a good PCB design takes time.

My wire-wrapping lately is a bit different, however. I start with blank, one-sided PCB stock, and CNC drill the stock for the through-hole pin patterns, then relieve the holes that do not connect to ground. If I were using SMT, then sure, PCB is the only way to go.

I'll attach a photo of what I'm talking about when I get to my camera.
 
Last edited:
You must log in or register to reply here.
Back
Top