RJ's Attempt. Maybe he's going to finish this time!

[rubber_jonnie]

For some reason the C input of the LS11 just wasn't making a connection to the pad, lord knows why, so it was bodge wire time.

I've had no end of odd soldering issues like that.. I half think to wipe the whole board with IPA before soldering.. But it doesn't seem to help. After talks and research, it seems the "hot air leveling" finish on the PCBs is a issue. Basically because the solder isn't perfectly flat.

Some of the plating seemed a bit bumpy under a magnifying glass, so this makes sense. Not fun with tiny little chips like the LS11 though. Thank god for bodge wires :)

[exxos]

Some of the plating seemed a bit bumpy under a magnifying glass, so this makes sense. Not fun with tiny little chips like the LS11 though. Thank god for bodge wires :)

Yeah, I mean I changed a videl on a Falcon last year without any issues, so everything else should be simple in comparison, but its never have been. I think the "best" method would be to solder the SMT pads first, then suck up the solder with wick and then solder paste it for the SMT stuff.. but thats a huge faff if theres a lot of SMT stuff. 2 fab houses have basically told me HASL finish isn't good for fine pitch SMT stuff... Even going back to my 4MB RAM boards, I had loads of issues there and a lot just refused to work, of course long story, but doing batch work, I think its probably more cost effective to reduce the failure rate by using ENIG finish.

Going back to failure on those plug in STE boosters, even soldering the CPU directly was like 90% failure rate. If it burned the board to a crisp the soldering was generally OK after that.. So I think any SMT stuff from now on, will have to be the ENIG finish else its going to be a world of pain and failures else.

Anyway, well done in the H4 and Mega fixes these past weeks! :cheers:

[rubber_jonnie]

Some of the plating seemed a bit bumpy under a magnifying glass, so this makes sense. Not fun with tiny little chips like the LS11 though. Thank god for bodge wires :)

Yeah, I mean I changed a videl on a Falcon last year without any issues, so everything else should be simple in comparison, but its never have been. I think the "best" method would be to solder the SMT pads first, then suck up the solder with wick and then solder paste it for the SMT stuff.. but thats a huge faff if theres a lot of SMT stuff. 2 fab houses have basically told me HASL finish isn't good for fine pitch SMT stuff... Even going back to my 4MB RAM boards, I had loads of issues there and a lot just refused to work, of course long story, but doing batch work, I think its probably more cost effective to reduce the failure rate by using ENIG finish.

Going back to failure on those plug in STE boosters, even soldering the CPU directly was like 90% failure rate. If it burned the board to a crisp the soldering was generally OK after that.. So I think any SMT stuff from now on, will have to be the ENIG finish else its going to be a world of pain and failures else.

Anyway, well done in the H4 and Mega fixes these past weeks! :cheers:

Thanks :) and whilst it's been quite hard work with the Megas and H4, everybody has given great support and advice, and it just shows what an awesome community we have here.

I guess it's tough, if HASL will work for SMT pre-assmbled boards then I suppose it's a good compromise, but then I guess that adds cost for assembly, so you may as well go for ENIG. I guess ultimately it's a call based on overall cost.

@exxos On another subject entirely, I'm trying to get some 27C4096 chips, the last lot I had were 55ns, but everywhere either has them on back order, or only has slower. What is the slowest you'd use for the H4 or a boosted 16/32Mhz machine? I can get 90ns but wondering if they might be a bit slow for anything boosted?

[exxos]

@exxos On another subject entirely, I'm trying to get some 27C4096 chips, the last lot I had were 55ns, but everywhere either has them on back order, or only has slower. What is the slowest you'd use for the H4 or a boosted 16/32Mhz machine? I can get 90ns but wondering if they might be a bit slow for anything boosted?

Stock machine any speed will do.. Though 55ns was pushing it for 32MHz on the STE.. 16MHz 90ns is pushing it, but should be doable..

Yeah, stuff is hard to get, I bulk brought the ROMs a few months ago among other things.. I keep saying Farnell is getting worse.. I been ordering stuff more and more from mouser lately as even when Farnell say they have 1,000's in stock, magically a few hours after I order stuff, its all out of stock and on back order for 6 months.. happend with the PSU caps many times, I gave up and brought a full reel of the things in the end. I just cant keep messing about with Farnell, if i order all the parts for some project to go off to be assembled and they go out of stock on just 1 item, it holds up the entire project for weeks.. then I have to spend time looking for other suppliers.l. time I don't have.. might as well throw in Farnell never learn to stop messing with the website as well :roll:

[rubber_jonnie]

@exxos thanks for that, I'll probably just get a couple of 90ns ones from Farnell, then I can use those for standard and use your faster ones elsewhere.

It's bloody frustrating,I'd use Mouser or Digikey, but I object to paying a minimum of £12 for FedEx for £10 worth of parts that could easily be put in the post in a jiffy bag.

Not only that, but they both all seem to have their stock in the US, and it ends up travelling halfway around the world. I'd rather wait and have it through a UK company, I don't know how these companies get away with not having UK stock, driving me mad.

[exxos]

Yeah. I think most of my parts come from overseas now :roll: only good UK supplier is rapid now..of course they don't have the range, but they don't mess you about. RS seems to be getting expensive and the amount of packing puts me off.

[rubber_jonnie]

Yeah. I think most of my parts come from overseas now :roll: only good UK supplier is rapid now..of course they don't have the range, but they don't mess you about. RS seems to be getting expensive and the amount of packing puts me off.

Rapid are just up the road for me, but like you said, they lack the range, nothing fancy like 27c4096's, which is a shame, though I could get other stuff from them.

[rubber_jonnie]

I was asked the other day for a Gembench test on my H4, so this morning I got a stock STF with TOS 1.04 and 1MB RAM, ran the full suite of GB tests and saved the results for comaparison against the H4.

The H4 differs only in that it has 4MB RAM. Neither had a Blitter installed for the test. I ran GB from a real floppy on both machines.

Here's the results:

H4_GB.jpg

Can't say fairer than that, the H4 matches the 1040STF absolutely identically. So far the H4 hasn't missed a beat with anything normal I've thrown at it. I also borrowed a Blitter from one of my Megas and dropped it in and used the diag cart to run the service tests and everything came back good. It's looking like a very happy machine this far.

I also tested HiRes mode and it all seems just fine.

I also tried SMonsons DVI adapter whilst I had a temporarily clear workbench, and the results weren't quite as expected, as below:

Low Res:

H4_LoRes_DVI.jpg

Med Res:

H4_MedRes_DVI.jpg

Although my dispaly does have some annoying problems, especially in Med Res it can be a bit fuzzy (Modern display), these are not the same as above with missing lines.I need to re-read the docs around the DVI adapter see if I've missed anything, and also test on another machine so I can verify it's working ok.

[exxos]

Odd.. Maybe a couple bad connections on the data bits ? @Smonson

[Smonson]

Odd.. Maybe a couple bad connections on the data bits ? @Smonson

I agree. Could you try running your finger around the back of the pin headers on the top of the blue board while it's running? It can sometimes change the colour of the dodgy bits if there are floating data lines (ground yourself on the chassis first to avoid static electricity).

I ran that board for an hour before sending it, but you can never be 100% certain with solder joints.

Was it working fine on a non-H4 machine before this? It could be a timing difference I guess, but it seems pretty unlikely.