Rockwell R6551A vs CDP65C51AE4

[jgroth]

Hi all,
I've got a bit of a conundrum here. My breadboard SBC (WDC65C02, WDC65C22, RAM, ROM and a modified version of Daryl Rictor's SBC-2 v2.5 decoding logic) is working fine with the Rockwell but the CDP either sends out garbage or in the case of CDP65C51AE2 complete silence. I bought all chips on Ebay so they could be faulty but I must have been very unlucky if all four ICs (two AE4 and two AE2) are faulty.
There is a crystal (1.8432 MHz) between pin 6 and 7. Pin 6 also has a cap (22pF) connected to ground.
DTR is not connected, DSR and DCD are grounded, RxClk is not connected and CS1 is connected to 5V, PHI2 (clock) comes from MPU PHI2 (pin 39), RTS and CTS goes to a MAX232.
Does anyone here know if there are any extra requirement for the CDP chip?

[GaBuZoMeu]

I've done a (very) quick scan of the CDP DS. I haven't seen anything special. If your Rockwell chip works, you may check whether you can see the receiver clock at RxC (I think if Rx clk src is set to internal then RxC acts as an output). With the same setting you might at least check if the BRosc of the CDPs is running correctly.
But I fear you have had really bad luck

[jgroth]

OK, will try that.

[EDIT] According to my oscilloscope there is a square wave coming out from RxCLX (pin 5) but it is not clean. The scope can't trigger the signal properly (or I might not know how to). I've read somewhere that some ACIA might need a 1MOhm resistor over the pins of the crystal, could that be it?

[jgroth]

So with some further testing I found out that both AE4 where dead (got only garbage out) and one AE2 was dead. The other AE2 which I thought was dead as well because it didn't send anything at all. It turned out it was not working because the /CS2 wire had crept out hence the chip was never selected. So one out of four ACIA is working, unfortunately not the one I wanted. Well, I'll try to find more 4MHz 65C51 somewhere, until then I'll see you on the bit stream.

[GARTHWILSON]

The reason for the 1M resistor across the crystal is to bias the input so the part will oscillate, and do so more or less symmetrically. If you're getting an RxC signal out, it sounds like it's oscillating. Hopefully you're well aware already, buy you need to be using the x10 position on the 'scope probe so you get correct results in your reading and don't foul up your circuit while you're probing it.

I've never had a bad IC of all the countless ones I've bought.

[GaBuZoMeu]

jgroth wrote:

I've read somewhere that some ACIA might need a 1MOhm resistor over the pins of the crystal, could that be it?

Yes, you can increase the chance to oscillate with a parallel resistor a bit, but it will affect frequency and stability as well, so it cuts both ways.

1 out of 4 is a bit disappointing - hope your financial loss wasn't too great

[jgroth]

GARTHWILSON wrote:

The reason for the 1M resistor across the crystal is to bias the input so the part will oscillate, and do so more or less symmetrically. If you're getting an RxC signal out, it sounds like it's oscillating. Hopefully you're well aware already, buy you need to be using the x10 position on the 'scope probe so you get correct results in your reading and don't foul up your circuit while you're probing it.

OK, so a 1M resistor is not needed as one the AE2 works like a charm. Will do some more testing with the AE4s. Just double checked the probe and yes x10 is selected.

[GARTHWILSON]

1M conducts enough to bias a CMOS input but not enough to interfere with the crystal working properly. I haven't needed it with Rockwell or GTE 65c51's. I don't remember if I've used any other brands.

[BigEd]

Getting a crystal to oscillate is quite a challenge - we had a thread about it - back in the day it was a value-add for a chip to contain the circuitry and to allow a crystal to be connected.

But these days a can which contains both the crystal and the support circuit is a better bet.

I see from a datasheet for a (hopefully) closely related part that you can either connect a crystal - with two caps, and that's a difference from what's described upthread - or you can inject a clock signal.

[jgroth]

BigEd wrote:

Getting a crystal to oscillate is quite a challenge - we had a thread about it - back in the day it was a value-add for a chip to contain the circuitry and to allow a crystal to be connected.

But these days a can which contains both the crystal and the support circuit is a better bet.

I see from a datasheet for a (hopefully) closely related part that you can either connect a crystal - with two caps, and that's a difference from what's described upthread - or you can inject a clock signal.

I do not have the second cap so will try that with the three other right now not working ACIAs.
Edit: The datasheet you provided the link for are Harris 65C51 of which I got two AE2s and one of them is working super fine. The other one must be dead because I can't understand how the /CS2 pin can be oscillating (logic probe confirms it and so do the 'scope).
I've got more hope for the AE4s. They output garbage right now but the /CS2 works fine so I wonder if the pin config is different. The company logo on the chip is a fancy 'i' so I wonder if that could be Intersil? Haven't found a 65C51 Intersil datasheet yet though.

[jgroth]

So I got one of the AE4s to work (haven't tested the other one yet and big thank you Gareth for giving me hope and not bin them). The AE4 needs the second cap which I hadn't installed. Once installed it works and is rock solid (so far ).
Which means virtual beers, wine, root beer, ginger beer or what ever you like to every one that helped me.
I do suspect the story will continue though, he he.

EDIT: Both AE4 work so the search for a 4 MHz ACIA is over.

[BigEd]

That's a good result!

I would expect the parts with same number but from different manufacturers to be highly compatible - that was the business, to have second sources and have independent suppliers. The different suffices would then indicate different packaging or different test results, or qualification for different conditions (like extreme temperature.)

In your case, it seems the AE2 parts pass the tests for 2MHz operation and the AE4 ones pass the tests for 4MHz operation. They might be identical designs which just happened to come out different, or the faster one might be a tweaked design.

So, I wouldn't expect different pinouts. But that doesn't explain what you've seen.

[jgroth]

You're are, of course, right BidEd. The pinout wasn't different, I was just grasping for straws. It was that second cap which I had missed reading the datasheet that was the culprit. Once installed, three of the ACIA work but one of the AE2 is brown bread. I can't get it to work but that doesn't matter as the two chip I want to use (the AE4) work just fine.

[BigEd]

Sorry, didn't intend to be bombastic - typing in a hurry!

[jgroth]

BigEd wrote:

Sorry, didn't intend to be bombastic - typing in a hurry!

It's me that should be sorry . What I meant is that I'm taking of my hat off for you, because you were right and I'm still learning. It usually pays off in the long run to listen to your teachers .