Atari XF551 floppy driver / power switch

[Dropcheck]

8 hours ago, macsonny said:

I doubled checked the soldering today and also reflowed a few points. Swapped the ROMs for stock Atari v7.4, Single patch and Hyper+ which by the way all work in my current XF551 fine. Doubled checked the 8040 and WD1772PH and also moved jumper to try different combinations. I finally have changed the ID settings but I'm not 100% sure I have the right ID selector switch. Ordered a new one from MOUSER per your par number just to be sure.

 

The floppy drive is a working mechanism as tested in XF551 fine.

 

I've not populated there track display components as figured they aren't critical at this point.

 

Swapped SIO cables just to be sure also. Same results.

 

Now I'm a bit stuck. 

Did you ohm out the floppy data drive cable to verify all lines are connected end to end?

[macsonny]

15 minutes ago, Dropcheck said:

Did you ohm out the floppy data drive cable to verify all lines are connected end to end?

I'm waiting on a new floppy drive cable to arrive as that's the next thing I was thinking might be the issue too.

 

Will report back in a few days.

 

[Dropcheck]

5 minutes ago, macsonny said:

I'm waiting on a new floppy drive cable to arrive as that's the next thing I was thinking might be the issue too.

 

Will report back in a few days.

 

If you have a multimeter, you should be able to confirm within half an hour if any internal lines are broken or mismatched.  😀

 

One other thought, did you substitute Q1 - Q4 transistors with another PN than the one in the BOM?  I have heard you can use 2N3904 and 2N3906, but I have not tested that since they have slightly different electrical characteristics than the ones in the BOM.  

[macsonny]

8 hours ago, Dropcheck said:

If you have a multimeter, you should be able to confirm within half an hour if any internal lines are broken or mismatched.  😀

 

One other thought, did you substitute Q1 - Q4 transistors with another PN than the one in the BOM?  I have heard you can use 2N3904 and 2N3906, but I have not tested that since they have slightly different electrical characteristics than the ones in the BOM.  

No, I used the transistors in the BOM. Should I replace them?

 

[Dropcheck]

2 hours ago, macsonny said:

No, I used the transistors in the BOM. Should I replace them?

 

I'm assuming that they are good still. 

[+DjayBee]

On 4/1/2024 at 1:49 PM, Dropcheck said:

The design does not take into account using a bigger eprom size than 2764, ie two distinct XF551 OSs switchable by the A12 line using the JP2 jumper.  So pin 26 is NC per design.  If you want to use a 27128 or bigger, you will need to bend pin 26/1 out of socket and use the normal hack to allow for the bigger size eproms.  Actually pin 1 is the VPP pin.

The idea was not having a switchable OS but to be able to use a 27128 in case one does not have a 2764 at hand.

Tying Pin 26 to a constant level would be enough.

[Dropcheck]

1 hour ago, DjayBee said:

The idea was not having a switchable OS but to be able to use a 27128 in case one does not have a 2764 at hand.

Tying Pin 26 to a constant level would be enough.

2764 eproms are still readily available for purchase on Ebay and elsewhere.  😔 

 

If the builder chooses to substitute with a completely different part number than what the BOM calls for, then he/she are effectively changing the design.  At that point the builder accepts the responsibility for figuring out whether that part number can work and/or what it needs to operate properly and make the necessary modifications.  I posted the schematic so the builder could.  I reject the premise I'm responsible to design for whatever part number the builder happens to have laying around.  That's just literally impossible to do.

 

If I had tied pin 26 to either GND or PWR, someone wanting to use a 27128 for 4 way switchable OS options would have had to cut that trace, in addition to wiring for a third/fourth OS.  In this case the substitution was the cause of the problem (wrong part number), not the design.

 

[macsonny]

On 4/3/2024 at 7:09 AM, Dropcheck said:

2764 eproms are still readily available for purchase on Ebay and elsewhere.  😔 

 

If the builder chooses to substitute with a completely different part number than what the BOM calls for, then he/she are effectively changing the design.  At that point the builder accepts the responsibility for figuring out whether that part number can work and/or what it needs to operate properly and make the necessary modifications.  I posted the schematic so the builder could.  I reject the premise I'm responsible to design for whatever part number the builder happens to have laying around.  That's just literally impossible to do.

 

If I had tied pin 26 to either GND or PWR, someone wanting to use a 27128 for 4 way switchable OS options would have had to cut that trace, in addition to wiring for a third/fourth OS.  In this case the substitution was the cause of the problem (wrong part number), not the design.

 

Update time.

 

So I received the ID selector switches per the BOM. I was a little concerned the ones I had weren't per BOM so got the right ones.

 

Bad news is after swapping out it's made zero difference.

 

Have you had a chance to look over the design and make sure you are sure there are no other errors in the PCB design?

[Dropcheck]

Well, rats.

 

I've gotten notice that the redo of the board pcbs should arrive from JLCPCB Monday or Tuesday.  So by the end of the week I should have more info.  I've done one w/o the optional control and track display and one with.  The only thing I could see is if I got the DataIN/DataOUT signals mixed up.  So I put both on jumpers with these new boards to confirm one way or the other.  I don't think I did. 

[_The Doctor__]

when in doubt, breadboard it, at the very least the unsure of or problem area

Edited April 7 by _The Doctor__

[macsonny]

On 4/7/2024 at 10:01 AM, Dropcheck said:

Well, rats.

 

I've gotten notice that the redo of the board pcbs should arrive from JLCPCB Monday or Tuesday.  So by the end of the week I should have more info.  I've done one w/o the optional control and track display and one with.  The only thing I could see is if I got the DataIN/DataOUT signals mixed up.  So I put both on jumpers with these new boards to confirm one way or the other.  I don't think I did. 

It seems someone has made some progress and found that Q1-Q3 are in the wrong orientation and should be spun around. Does that seem correct to you?

[Dropcheck]

2 hours ago, macsonny said:

It seems someone has made some progress and found that Q1-Q3 are in the wrong orientation and should be spun around. Does that seem correct to you?

It depends on a number of facts and circumstances.  The truth is in the details.  Who and what and where is this info coming from?

Edited April 8 by Dropcheck
Added info

[+CharlieChaplin]

1 hour ago, Dropcheck said:

It depends on a number of facts and circumstances.  The truth is in the details.  Who and what and where is this info coming from?

 

Abbuc floppy doc - he has a lot of experience with A8 floppy drives. And he found more bugs on the re-imaged XF pcb's 3E and 6.0...    https://abbuc.de/forum/viewtopic.php?f=3&t=2763

(use Google translate or DeepL to translate from german into english)

 

[Dropcheck]

The only 'bug' in the XF551 v3E appears to be the often misunderstood CPU jumper setting.  It was actually something that I didn't fully understand until recently.  For standard 8040 and 8050 CPUs not coming from Atari, the id setting should be 8040.  If you happen to have the pre-programmed 8050 cpu that escrewed the ROM chip on some later produced XF551s then you would need to set the 8050 id.  For the longest time I thought and many others thought it was simply related to the part number of the CPU.   That sometimes caused a backwards setting of the jumper.  It was puzzling, but not in any sense a major bug, since swapping the jumper solved the problem. 

 

On the XF551_B v6.0 I have found a definite error on my part pertaining to the power supply, that was caused when I added the additional options for a different power switch and power plug barrel.  As I redid the schematic, I left off a connection in the schematic.  My test board had worked normally with the original Atari parts for the switch and power jack, so I saw no reason to redo the board for parts that were almost direct dropin replacements except for a slight different footprint for retainer pins.

 

One of the 'errors' appears to have been caused when a completely different part number was substituted.   A 27128 eprom for the BOM listed 2764.  The design of the board had left pin 26 NC for U6.  It was shown on the schematic as such.  The 27128 required that pin 26 be connected either to GND or +5V.  I don't consider that an error on my part.  It's not  uncommon that people substitute parts, but I can only verify that it works with the parts I have listed in the BOM. 

 

The rest of the 'supposed errors' on the v6 board have not been verified by me.  Other than that power supply trace error, the schematic still appears to be correct.  I am looking into whether or not there was corruption in the gerbers production process when I redid the board for the previous mentioned optional power switch and jack.  It doesn't seem likely, but I've seen stranger things happen for no apparent reason. 

 

[_The Doctor__]

Nothing wrong with making the PCB as robust as it can be.

 

As for the 8040 / 8050 it's kind of a silk screen instruction issue. It might be that it should read..

80XX ext ROM /8050 int ROM

 

Clears things up a bit when looked at.

 

All the more reason to thank and build on each other.

[macsonny]

17 hours ago, Dropcheck said:

The only 'bug' in the XF551 v3E appears to be the often misunderstood CPU jumper setting.  It was actually something that I didn't fully understand until recently.  For standard 8040 and 8050 CPUs not coming from Atari, the id setting should be 8040.  If you happen to have the pre-programmed 8050 cpu that escrewed the ROM chip on some later produced XF551s then you would need to set the 8050 id.  For the longest time I thought and many others thought it was simply related to the part number of the CPU.   That sometimes caused a backwards setting of the jumper.  It was puzzling, but not in any sense a major bug, since swapping the jumper solved the problem. 

 

On the XF551_B v6.0 I have found a definite error on my part pertaining to the power supply, that was caused when I added the additional options for a different power switch and power plug barrel.  As I redid the schematic, I left off a connection in the schematic.  My test board had worked normally with the original Atari parts for the switch and power jack, so I saw no reason to redo the board for parts that were almost direct dropin replacements except for a slight different footprint for retainer pins.

 

One of the 'errors' appears to have been caused when a completely different part number was substituted.   A 27128 eprom for the BOM listed 2764.  The design of the board had left pin 26 NC for U6.  It was shown on the schematic as such.  The 27128 required that pin 26 be connected either to GND or +5V.  I don't consider that an error on my part.  It's not  uncommon that people substitute parts, but I can only verify that it works with the parts I have listed in the BOM. 

 

The rest of the 'supposed errors' on the v6 board have not been verified by me.  Other than that power supply trace error, the schematic still appears to be correct.  I am looking into whether or not there was corruption in the gerbers production process when I redid the board for the previous mentioned optional power switch and jack.  It doesn't seem likely, but I've seen stranger things happen for no apparent reason. 

 

Good news - sort of.

 

I flipped every single transistor (Q1-Q4) 180 degrees and now the floppy drive boots to the ROM (Hyper+) menu.

 

The floppy derive doesn't work when I try to boot but apparently the following needs to be done:

 

U2-3 -> U1-12
U2-4 -> U1-13
U2-5 -> U1-14
U2-6 -> U1-15
U2-7 -> U1-16
U2-8 -> U1-17

 

Means a few track cuts for U2 pins 3-8. I haven't tried this yet but I've been lead to believe it works. Will report back when done.

[Dropcheck]

2 hours ago, macsonny said:

Good news - sort of.

 

I flipped every single transistor (Q1-Q4) 180 degrees and now the floppy drive boots to the ROM (Hyper+) menu.

 

The floppy derive doesn't work when I try to boot but apparently the following needs to be done:

 

U2-3 -> U1-12
U2-4 -> U1-13
U2-5 -> U1-14
U2-6 -> U1-15
U2-7 -> U1-16
U2-8 -> U1-17

 

Means a few track cuts for U2 pins 3-8. I haven't tried this yet but I've been lead to believe it works. Will report back when done.

Hold up.  From what I've read the builder still doesn't have a functioning board.  There maybe other problems as well.

 

Redone boards came in yesterday afternoon.  I've started assembling the one without the optional track display and front controls. 

 

 

[Dropcheck]

The only issue I am concerning myself now with is whether there was a complete electronics CAD software failure in producing the gerbers for the XF551OriginalBv6.0 pcb.  It's looking more and more as if there is.   Is that my fault?  I have never tied multiple traces to wrong pins on the U1 and U6 and U2 chips before, ever.  The schematic is apparently correct, (I've been doing these boards for almost 10 years now), so the software should have routed per the schematic.  I have sometimes changed the footprint of the transistors, to accommodate the specific part number I chose to use, and for spacing needs.  A bad part pin to footprint hole assignment could have happened there.  That could be me.   But when you put it together with the major routing errors on the chip pins, it's less likely. 

 

Why the previously posted gerbers appear to be so mangled is a mystery at this point. The fully routed Diptrace .pcb board that was used to generate the gerbers is showing a correctly routed board.  The gerbers produced and uploaded to my website have the same date and very close time stamp to the date and time stamp of the fully routed Diptrace.pcb file.  So something happened between the conversion process and the pcb manufacturing.  With more than one person having the same issue, it points the finger at the file itself and whatever happened to it once it was created.  Did the corruption happen on my computer or did it happen when it was uploaded to my website, downloaded to the builder's computer and then sent to the pcb manufacturer?  How could the software take a correctly routed board and produce a set of gerber files that don't reflect the routing?  At this point I have no explanation that makes any sense.  

 

The missing power trace has already been addressed and I've accepted the fault there and explained why I feel that happened. 

 

I'm assembling new pcbs right now to test.  So far the pcb routing seems to be following the schematic.  Which it should.

[Dropcheck]

FYI:

 

First short board assembled sans the critical 8050, WD1772, eprom and OSC.  First power on check.  No explosive decompression.  LEDs lit.  I have +12V and +5V at the disk drive power connector.  I have +5V on the power rail of all chips.  So it appears the power supply circuit is good and checked off.  I fixed my known mistake.  

 

The next part may be either fast or slow depending on the will of the GODS(ie good 8050,WD1772, eprom and OSC) and the correctness of the board routing.  Hopefully the next step of doing a static check of the fully populated board without computer connection will pass as well.  Stay tuned. 

 

 

 

[Dropcheck]

Update:

 

Static test of XF551B_v6 short board has passed.  Drive spun up, drive light came on the front and stopped.  That is normal.  It appears that I have good OSC, WD1772, eprom and 8050. 

 

Now I just need to reconfigure my test bench and find some way to do a video of the next step, doing a full test of computer/disk drive SIO communication.   I haven't tried doing video on my phone or my Nikon camera yet, so that might not happen until later on this afternoon. 

 

 

 

[Dropcheck]

Update:

 

With this current board, I had absolutely no problem with assembly and testing.  It worked first time out without any issues that I could determine.  I have uploaded a Youtube video showing the testing.  Sorry for the wheezing in the audio.  I've been fighting bronchitis all winter long.  

 

XF551v6 Short Board

 

Based on what I have been able to determine so far.  There is a possibility that in changing the transistor footprint from one design to another, I may have mis-connected the symbol and footprint.  Two builders have reported that flipping the transistors 180 degrees corrected that and appears to confirm that possible cause.  The previous testing board was working, but I wanted to be able to simply drop in the transistor without having to bend/form the leads.  It should have made it easier for the builder too, assuming they used the exact part number called for in the BOM.  The same desire to have alternate parts for the power switch and power connector apparently caused me to drop a power connection trace.  That I absolutely own.  The transistor issue was probably my fault as well. 

 

As for the trace errors between U1 and U2, I have no clue how that happened.  None of the schematics I have from that time show those erroneous connections.  None of the routed Diptrace pcbs I have show that trace routing.   Without some evidence either in a schematic or pcb layout file I am unwilling to take responsibility.  It's obvious something happened between my creation of the gerbers, uploading to my websit and/or at the pcb manufacturer.  But at this point I don't think we are going to know how or why it happened.  You might have an opinion, but there is no evidence of my fault in this. It's not anything I did or didn't do.  It would have shown up in the pcb layout file first, and it didn't.

 

I still have a second board that includes the optional track display and front control circuitry that I will build up in the next couple of days.  If it passes as well, then I will go ahead and produce the replacement gerbers for upload to my website.  @Macsonny and the builder Erhard on ABBUC will have a free set of 5 bare boards sent to them as an apology for the massive failure of this particular board version.  It won't make up for the cost of the parts and time wasted, but at least if they ever want to try it again, they will have the boards on hand.   I don't have posting access on ABBUC at the present time, or I would contact Erhard myself.  If anyone can alert him to this post.  I will be trying to get ABBUC forum posting access as soon as possible.  

 

I am sorry for the mess and cost and time wasted. 

 

NOTICE:

 

Anyone wanting to build up any XF551 style pcbs, delete any files previously downloaded from my website, and then re-download them.   That should insure the latest and corrected gerbers and other documentation.

Edited April 11 by Dropcheck
added apology

[+DjayBee]

4 hours ago, Dropcheck said:

If anyone can alert him to this post.

I did.

 

If you PM me an email address, I'll forward it to Erhard.

[macsonny]

5 hours ago, Dropcheck said:

Update:

 

With this current board, I had absolutely no problem with assembly and testing.  It worked first time out without any issues that I could determine.  I have uploaded a Youtube video showing the testing.  Sorry for the wheezing in the audio.  I've been fighting bronchitis all winter long.  

 

XF551v6 Short Board

 

Based on what I have been able to determine so far.  There is a possibility that in changing the transistor footprint from one design to another, I may have mis-connected the symbol and footprint.  Two builders have reported that flipping the transistors 180 degrees corrected that and appears to confirm that possible cause.  The previous testing board was working, but I wanted to be able to simply drop in the transistor without having to bend/form the leads.  It should have made it easier for the builder too, assuming they used the exact part number called for in the BOM.  The same desire to have alternate parts for the power switch and power connector apparently caused me to drop a power connection trace.  That I absolutely own.  The transistor issue was probably my fault as well. 

 

As for the trace errors between U1 and U2, I have no clue how that happened.  None of the schematics I have from that time show those erroneous connections.  None of the routed Diptrace pcbs I have show that trace routing.   Without some evidence either in a schematic or pcb layout file I am unwilling to take responsibility.  It's obvious something happened between my creation of the gerbers, uploading to my websit and/or at the pcb manufacturer.  But at this point I don't think we are going to know how or why it happened.  You might have an opinion, but there is no evidence of my fault in this. It's not anything I did or didn't do.  It would have shown up in the pcb layout file first, and it didn't.

 

I still have a second board that includes the optional track display and front control circuitry that I will build up in the next couple of days.  If it passes as well, then I will go ahead and produce the replacement gerbers for upload to my website.  @Macsonny and the builder Erhard on ABBUC will have a free set of 5 bare boards sent to them as an apology for the massive failure of this particular board version.  It won't make up for the cost of the parts and time wasted, but at least if they ever want to try it again, they will have the boards on hand.   I don't have posting access on ABBUC at the present time, or I would contact Erhard myself.  If anyone can alert him to this post.  I will be trying to get ABBUC forum posting access as soon as possible.  

 

I am sorry for the mess and cost and time wasted. 

 

NOTICE:

 

Anyone wanting to build up any XF551 style pcbs, delete any files previously downloaded from my website, and then re-download them.   That should insure the latest and corrected gerbers and other documentation.

Wow - how amazingly generous and humble. I really didn't expect you to be that generous but also appreciate the offer. I sort of accepted that this was a bit of a trial-and-error scenario so was more focused on solving the problem. However - I really appreciate the gesture.

 

If it helps I really enjoy these types of projects so time is never wasted. It' all part of the (fun) experience for me. And as for parts - I'll have enough parts to build two boards now. Better get my disordering gun all fired up!

 

[+DjayBee]

One more thing:

https://abbuc.de/forum/viewtopic.php?p=22427#p22427

 

Quote

 

Yesterday I pulled T1 (that tiny little SMD transistor) off the board and measured it's pin assignment.

The pin assignment - guess what - does not match the pcb layout. The little beast is a PNP transistor and the pcb layout connects it:

+5V -> collector
/side -> emitter
LEDDOTS -> base

The correct assignment would be:

+5V -> emitter
/side -> base
LEDDOTS -> collector

I had to turn T1 like shown on the picture below and connect the base using a wire. Now the LED dots are on when side 1 is selected only.

 

 

[_The Doctor__]

definitely make sure the instructions to fix the messed version board go far and wide but none of it's creation files continue to propagate and ask others make them go bye bye. Anything is fixable of course . The new revision is here already though.

 

The last version 6 looks to be moving along nicely so it's all good. Happy times.