• Please review our updated Terms and Rules here

Diagnose issues: PET 4032 with PETTEST ROM

I keep thinking of Clint Eastwood in the Dirty Harry films saying "You've got to ask yourself one question: 'Do I feel lucky?' Well, do you, punk?"

Dave
 
Hi,

It is good having the same reference documentation for the machine - we stand a chance of talking the same language now we have a common frame of reference!

UA19 is responsible for data bit 0 (D0) of the lower bank of 16K DRAM. So this would be the chip to try. However, I am thinking that the data bit is stuck at '0' - and this (generally) is not good for a piggy-back test to work. Stuck at '1' is usually much better.

The above is if it is a data bit fault.

You could also have an address fault - in which case, piggybacking UE8, UE9 and UE10 in order could yield some results.

Obviously the machine is working enough to run my PETTESTER - so that rules out quite a lot of issues before we start.

On the oscilloscope front - @Hugo Holden has recommended a Hitachi V-509 oscilloscope - and I will tell you how good that is when I open my parcel from the post man presently :)!

Dave
 
daver2 said:
On the oscilloscope front - @Hugo Holden has recommended a Hitachi V-509 oscilloscope - and I will tell you how good that is when I open my parcel from the post man presently :)!

Dave
Click to expand...

Lucky you ! I think you will be surprised how compact these units are.

You should have lots of fun playing with the dual delay timebase. A good place to start is to examine a standard composite video signal using the internal V sync separator. These scopes have very effective sync separators. Then when you have the main timebase showing one field (two vertical syncs) on each side of the trace, switch to intensify and adjust the vernier and the B timebase knob to create the window of the lines you want to inspect, or the V blanking and sync areas if you like, and switch from intensify to B to display them. Using the vernier you could inspect every line too.
 
I'd second the clean up idea, while a dead 4116 RAM is the most likely cause (they're quite well known for random failures at this point), it's possible that you have a bad address line contact on the CPU socket (some good contact cleaner helps there), could have a damaged trace hidden under the dirt, or even possibly some dirt being conductive enough to cause weird issues.

Lots of things can cause address/data lines to be stuck in a state, including other defective ICs such as the PIA and VIA chips, a defective ROM chip, or of course defective ram. It's also possible that the bad chip isn't even in the first banks of ram but in one of the higher ones but with it's output stuck on all the time so it's fighting with other chips.
 
OK, I cleaned up the board with some isopropyl alcohol and some q-tips. Unfortunately, no change on the stuck screen. What is my next step? If I get an oscilloscope, will it be easier to find the bad chip(s)?
 
Hello, friendly ping on this thread. in your opinion, is it better to buy an oscilloscope to test for bad chips or buy a chip tester and start removing them (replacing with sockets)?
 
irataxy said:
Hello, friendly ping on this thread. in your opinion, is it better to buy an oscilloscope to test for bad chips or buy a chip tester and start removing them (replacing with sockets)?
Click to expand...
I think most would agree the way to go is the scope. Otherwise you would find yourself removing a lot of good/normal vintage IC's for no reason. Also there is a risk to the pcb unless your de-soldering tools and soldering equipment is excellent and sometimes this can introduce new faults. Also it is better to keep as many of the IC's as original as possible with their date codes etc matching the computer.

Ideally an IC should only be removed when there is conclusive evidence it is defective; meaning that its input and output pins have been checked (typically with the scope) and it is found not to be obeying the IC's logic table. I find it is better to have the TTL data book on hand to be able to quickly look up the logic table and pin outs for most 74 series types, or look up their data sheets on the net.

www.ebay.com

The TTL Data Book for Design Engineers, Texas Instruments 1981, Hard Cvr. 2nd Ed | eBay

1981 The TTL Data Book for Design Engineers Second Edition.
www.ebay.com

Mostly (there are exceptions) the faults in TTL IC's result in their output stages failing to operate or having defective logic levels. So this can be detected as abnormal voltages on their output pins. It is rare that the inputs fail drawing current loading another IC's outputs, but on rare occasions inputs can go open circuit inside the IC assuming a high logic state in there, giving unexpected logic output states. (on super rare rare occasions inputs can short together, but I think this has only ever been seen once on this forum)

I'm just proof reading a final draft of an article on finding defective 4116 DRAMs in the 2001 series Dynamic PET's, I will post it this week. There will be some info in there that will help if you have DRAM problems. There are at least 4 ways the DRAM's can behave when defective and there can be problems in their support circuitry. Unlike SRAM, the DRAMs require a moderate amount of additional support circuitry to control them properly and generate the correct row & column address sequencing for them. For example the 4116 has 7 address input lines and it is critically dependent on the /RAS and /CAS (row & column address strobe signals) to generate the address locations (effectively a 14 bit number) inside the IC, to specify one of the 16384 memory cells. These IC's are very complex with internal clocks. Also the /RAS is important for refreshing. SRAM does not have these complications. The method I have deployed is to use SRAM to support the computer operations so the 4116's can be examined in detail with test programs in added ROM and I have made 24 scope recordings of the important signals in the DRAM control system.

As for testing chips with "chip testers" (sounds great on the face of it) once removed from the pcb, this is straightforward for most 74 series logic chips. However, much less easy for the 4116 DRAM I have found. For example, I bought a 4116 DRAM chip tester and it reported that twenty NOS 4116-15nl IC's were defective, when they are not and work perfectly in the computer. This faster part is somehow confusing the chip tester. So after this I have put much more faith in testing the 4116 IC's in the actual computer where the designers have configured the control circuitry to work correctly. I have bought a second 4116 tester of a different design and I'm currently evaluating this one.
 
+1 - Don't start desoldering stuff needlessly (for the reasons Hugo has already stated).

We already know we have a large amount of the PET working - as it runs my PETTESTER... This automatically rules out lots of things (most of which is mentioned in post #25).

There are some other tools available - but they involve having a 'digital twin' (i.e. another identical machine) available to compare results with.

Dave
 
Wow, great info @Hugo Holden !! Thank you.

I'd like to buy an oscilloscope in advance of the article you spoke of. Can you recommend a low-cost (<$100), beginner-friendly scope? It only needs the ability to diagnose problems in the PET, nothing new or sophisticated (I'll upgrade later if needed).
 
Hugo Holden said:
For example, I bought a 4116 DRAM chip tester and it reported that twenty NOS 4116-15nl IC's were defective, when they are not and work perfectly in the computer.
Click to expand...

It wasn't one of those ones based on an Arduino Nano, is it?

I actually built one of those myself (well, not the whole thing, I just wired up a breadboard and drove it through the USB connection, no buttons or whatever) and while it "worked" I came to the conclusion that at least the code that was being passed around at the time was probably hopelessly not-like-a-real-world application. Maybe someone's written a better version using assembly by now; the simplistic Arduino C code I was playing with was slow enough that I suspect it was substantially outside the refresh recommendations in the datasheet.

(I think you could do a perfectly good one with an Arduino nano, with assembly you could probably run it at close to an effective 4mhz or so.)
 
irataxy said:
Wow, great info @Hugo Holden !! Thank you.

I'd like to buy an oscilloscope in advance of the article you spoke of. Can you recommend a low-cost (<$100), beginner-friendly scope? It only needs the ability to diagnose problems in the PET, nothing new or sophisticated (I'll upgrade later if needed).
Click to expand...
I generally recommend the small Hitachi analog scopes as starters, V209 or V509 because they are super compact and don't occupy much bench space at all. The 509 has a delay time-base and is 50Mhz, the 209 is 20 MHz, both would work to help fault find a PET. The 209 is often cheaper and plentiful and at least close to $100 and infinitely better than any new digital scope at that price area. One feature is they have a very high final anode voltage for the small sized CRT at 10kv and they have a super bright trace.

The trick is, never buy one that is cosmetically beaten up, missing its carry handle. If you can find a good looking one cosmetically & working, invariably it will have been looked after. The V209 appears on ebay frequently:

www.ebay.com

Hitachi Oscilloscope V-209 20MHz | eBay

Find many great new & used options and get the best deals for Hitachi Oscilloscope V-209 20MHz at the best online prices at eBay! Free shipping for many products!
www.ebay.com
 
Eudimorphodon said:
It wasn't one of those ones based on an Arduino Nano, is it?
Click to expand...
I don't think it is an Arduino, I will check.

In the meantime I have found it also reports the occasional 4116-3 IC as defective too, when it is not. I guess these could be IC specimens that run faster in that case. It must be a timing issue.

I bought another tester on ebay that appears to use a small OLED display (or similar). It tests all 4116 DRAM's correctly and the ones it deems defective, are also defective in the computer. And the ones it says are good, work in the computer ...so far.

It is a lesson though, about testing complex parts in an alternate testing device, vs the application they are used in. The test process could have false negative or false positive results (just like any medical test). If a positive is defined as a dud IC, then the first tester I got has significant false positives. And in the case of false negatives, the DRAM could be bad when the tester says its good. Most likely this would come about with refresh problems, if the tester tested the IC over too short a time frame. Without refresh, depending on the IC, some can store charge on the capacitors for many minutes, but in fault conditions it can be much shorter and the capacitor/s in a particular cell can lose their charge sooner which may or may not show up with rapid testing in a tester.

( I had some fun & games in my workshop last night. My 2465B scope failed in front of my eyes, the H deflection failed. In any case I have spare 2465B scopes in my workshop immediately deployed another that had not been used for about a year. Within a few minutes it emitted a huge quantity of smoke filling the workshop. The Rifa filter capacitors on the line side failed as they often do when aged. So I stayed up late fixing that and now I'm working on the other failed scope. Luckily I collected a very large amount of spares for the 2465B, but still, it is upsetting when a TEk scope breaks down)
 
Last edited:
Hugo Holden said:
Luckily I collected a very large amount of spares for the 2465B, but still, it is upsetting when a TEk scope breaks down)
Click to expand...

I dread the day my beloved Tektronix 453 goes down. I have had it for 20 years and only had to replace the power cord. I noticed that it had a National Cash Register label so no telling how many hours of service it has.
 
Hi Irataxy.

Two oddball thoughts, maybe useful maybe not. Feel free to ignore me.

1) Does your ROMulan let you override the onboard RAM? I have a little board I got from a fella on this forum (Nivag) that does both RAM and ROM. The RAM is not selectable IIRC, just fully on or fully off. If you did that, you might be able to see a working machine at least, and likely narrow down the issue to a bad RAM chip. The issue could still persist if your display RAM is stuck (the daughterboard I got does not replace the video RAM).

Is the video RAM on here at UC4 and UC5?

2) Adrian Black did a couple videos on some cheap O-scopes recently. There was a hantek 6022BE, that's software based, which is why its so cheap ($67 on Amazon today). Use open hantek and not the vendor software if you get it.

The folks here are right. Scoping the data and address lines will be super helpful in diagnosing the issue.

Good luck!
 
My PETTESTER firmware does a preliminary test of the video RAM before it moves on to test page 0 and 1 RAM, which is where the test is stuck.

To be effective, you need a NOP generator to scope the address lines and address selection logic. Some ROMulators have a NOP generator built in.

You need to be careful with 'cheap' software oscilloscopes to make sure they can adequately measure the clock frequencies that exist on a PET. Cheap ones can't...

Dave
 
I successfully used a Hantek 6022BE to repair my PET and got all the clock frequencies just fine. The supplied software worked ok for me too, but OpenHantek is more flexible (there's a Linux version for example).

Colin.
 
If you suspect a digital issue, a very handy tool for quick diagnosis and tracing is a simple logic probe. You give it power and ground and you can poke it into the circuit on IC legs and such and check if the signal is high, low, floating (nothing driving it), or fluctuating (either rapidly or pulses). They're honestly much faster to work with than an oscilloscope for certain tasks, albeit more limited, and much much cheaper. Unlike a logic analyzer there's only one channel but they're much more responsive for quick checking if signals make sense in a digital circuit.
 
With the caveat that their frequency response is severely limited (when compared to a reasonable oscilloscope). You have to be careful what signal you are therefore trying to measure.

Dave
 
You must log in or register to reply here.
Back
Top