I guess I missed that part, him having the Rockwell unit. It's tough to get old and not be able to see straight.

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

Then those published circuits are wrong. The MAX232 and 238 have used 1uF charge pump capacitors since the early 1990s, when the chip underwent redesign. Again, please review the data sheet.

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

Before you guys have a mini bust-up over capacitor values, I'd just like to say I have acquired a set of 1uF electrolytics this afternoon, as well as a new MAX232 just in case the old one is blown.

I also got a new multimeter which has a frequency counter (I am hoping I can see the clock signals with it).

No bust-up. Just don't want anyone getting messed up by published misinformation. Incidentally, suffixes like CPE on the MAX parts refer to the packaging and temperature rating, not the electrical characteristics. For example, CPE refers to a PDIP package with a 0°C to +70°C temperature range, ESE a narrow SO package with a -40°C to +85°C temperature range, etc. To order a MAX232 in a PDIP package (16 pins) with a commercial temperature range, you'd ask for a MAX232CPE. The latest data sheet is right here.


That'll help, but you should also get your hands on a good logic probe. Later on and if finances allow, look around for a used 'scope if you are going to develop more ambitious designs. Being able to see waveforms can sometimes assist in troubleshooting recalcitrant circuits. However, the logic probe will be your most useful tool. I have an old B-K DP-21 model I purchased around 1990, which is effective to 20 MHz and works with TTL and CMOS. If it were possible to wear out a logic probe, this one would have been in a landfill long ago.

_________________
x86?  We ain't got no x86.  We don't NEED no stinking x86!

Well..

The meter is picking up 50hz from the power supply (even with just the -ve probe touching gnd), and I noticed this with the piezo transducer. Power supply is switched mode DC 5v 2A rating. Measures 5.1v DC but near zero AC.

There is nothing on any clock line apart from 50Hz mains hum
and nothing on the crystal - I expected 8Mhz.

Don't know if this meter is able to measure square wave frequency or whether it is AC only - Manual doesn't say.

Logic probe is in the post, it is one of these

http://www.maplin.co.uk/logic-probe-567 ... &u=strat15

This will show logic High / Low and pulsing (ie clock). I've yet to understand what this gives that a multimeter doesn't, apart from knowing a line is pulsing (but then you don't know the frequency). Still for a tenner it will be easier to use than a meter.

They undoubtedly went to a higher switching frequency so they could reduce the size of the capacitor to make the product more attractive. If you ask, they probably give you the conservative side anytime the person you're talking to is not totally sure. Looking at other switched-capacitor voltage converter datasheets, I don't find any upper limit to the capacitor size, except that they say that going beyond a certain value for a given switching frequency won't give any benefit. The story, as I remember it from one of the industry trade magazines, is that the MAX232 spawned a gob of switched-capacitor voltage converters because it was found that a lot of customers were using it not for RS-232 but for power conversion. I don't know when the 7660 came along, but there are a lot of variations on that, some of which I've used (from Linear Technologies, National, Intersil, and Maxim), variations for higher current, higher voltage, even regulated, etc..


Leakage of electrolytics is quite minimal. I've kept memory alive on a 10uF for a day (didn't try any longer), and a product we had 20+ years ago charged a 220uF 'lytic through a 1M resistor to get a 220-second time constant, and it was consistently and surprisingly pretty accurate, even without using precision or low-leakage capacitors, although it was at temperatures that were comfortable to humans (ie, we didn't have to run it hot).

Tantalums will have lower ESR. If you can get OS-CON capacitors, that's even better. OS-CON is not a brand but an electrolyte type, and there are at least two companies making OS-CON capacitors, one being Sanyo (which we use for the small switching supplies in our products).


What's the upper limit? My DMM has an upper limit of 200kHz on the frequency counter, and I can hardly imagine connecting the long leads to a MHz or higher square wave. (I'm not saying it can't work for that, but I'll be interested.) After I wrote this, I see you added:

I expect it is AC-coupled, so the DC component shouldn't matter as long as it's not so high that it blows the input capacitor (like a few hundred volts).


A 0-5V square wave with a 50% duty cycle would probably just appear as 2.5V on a multimeter, as if it were sitting there quietly as a DC voltage. Now you'll know it's actually pulsing between two valid logic states.

If you already had a logic probe, as BDD has, you might not see much extra value in a frequency-counting multimeter. Especially if you already had a scope. But if you already had a frequency-counting multimeter (as I have) you might not see much extra value in a logic probe. It does tell you something about mark/space ratio, I gather, but then so does the mean DC voltage perhaps?

(Frequency counting up to 15MHz)

It is one of these:

http://www.maplin.co.uk/ut60e-true-rms- ... face-46458

Goes up to 10Mhz for frequency counting. I needed a new meter anyway as my old one (30 years and still working) is showing signs of intermittent failure.

Question is can it measure square wave frequencies? Because I am still not convinced the clock circuit on my board is not functioning, and if the meter can't do this one thing it is going back where it came from.

Well, mine can measure clock frequencies up to 8MHz and beyond, so I think you can conclude that your clock isn't a clock.

Darn it..... It's the simple answer isn't it?

can you rig up a clock from any other nearby source? Or do you need to send off for another crystal? One reason I like the Beeb as a platform for experiments is the ease of picking signals out of it, or desocketing chips. If you've any old 8-bit computers around, you might have some signals of reasonable frequency.

Good idea. I've a ZX Spectrum+ 48K here... when I get a chance I'll check the CPU's clock to see if the meter can detect it.

Is there a way to test the crystal independently, and are they fragile? I mean, what's the chance I damaged it during soldering?

Sorry, I don't know how fragile crystals are. I think perhaps (4-pin) crystal cans are more robust but I get the impression this really is a (2-pin) crystal. (I think it's a fairly good plan to socket as much as you can, and perhaps a moderately good plan to buy a spare, for components which are cheap compared to postage. If you use a socket then you can't damage during soldering, and you can replace a component which got damaged some other way.)

But, just a quick and obvious question: Have you checked your board for shorts and opens? And checked the power rails, of course.

(Edit: I should point out: the 4-pin crystal can is a 5v oscillator, and produces a clock signal without the need for the external components found on the micro-uk101)

There isn't a power rail as such, the Vcc / Gnd lines are all connected pin to pin. Yes, they're all correct and supplying +-5v to Vcc / Vss/Gnd respectively at all required locations (basically the IC sockets). I have done continuity checks on the IC pins to ensure all address and bus lines are connected, and the glue logic connections too. Like I said, i found a few errors which I corrected. I found incorrect resistor values in the clock circuit and corrected them (47k instead of 470R). I think I should revisit the wiring of the clock circuit itself to double check it. I already checked it once.

Yes, it is a 2 pin crystal. Only the ICs are socketed (otherwise I may just have well have used a breadboard).

I think I have a couple of spectrums, but can only find one, and it doesn't seem to have any life. There's a schematic herewhich says the ULA clock output comes out on pin 8A of the edge connector, and the board diagram here suggests that pin 8A would be on the underside, and would count from the middle, towards the power connector. It might be that A0 would also be an acceptable clock-like signal - pin 9A. Oh, in fact there's a handy diagram of the edge connector which I missed: