Yamaha O1V PSU conundrum.

[Sgt_Baker]

TL;DR: Has the whole PSU/Transformer subsystem failed?  Crackling transformer even when disconnected from AC -> DC board.  All "digital brain" still works.

Hi all,

I'm the proud owner of a classic car Yamaha O1V mixer - S/N 001234 (not making that up) - which has become rather unwell of late.  Honesty factor: While I'm nowhere new to electrical diagnostic and repair, it certainly isn't close to my full-time job, nor do I possess any more sophisticated diagnostic tools than three crappy multimeters, ears and nose and an oscilloscope so ancient it's branded as "Post Office Telecommunications" (the UK's national telecoms agency prior to British Telecom (Bastard Telecom?) coming in to existence).

So, the Yamaha PSU.  Scroll to the very last page of this document: https://ultramfcd.com/img/YamahaO1V/O1V_service_manual.pdf

Early O1V models were known to exhibit a grounding fault, whereby the AC -> DC board was split in to three different ground planes reliant on physical screws to the DC PSU sub-chassis for mutual grounding.  Yamaha's own research indicated that "mishandling" of the shipping package during transit might cause one or more of said grounding screws to work loose, de-referencing one or two of the PSU's ground planes.  Indeed, early symptoms - display ceasing to work and the occasional crash - could be cured with a swift tap to the chassis.  The solution (as per Internet) to this problem is to bodge the three ground planes together thus:



Sorted?  NO!  Once reassembled the -15V (audio) regulator immediately spilled its innards and cast smoky assertions across the whole board.  For reference, here's a screen capture from a vintage scope of the -15v rail during power-on:

https://ultramfcd.com/img/YamahaO1V/SwitchOn.mp4  That's a wild +/- 5V oscillation.

So got pissed off.  Now I've replaced the +/- 15V regulators and it transpires the transformer is still making weird and worrying cracking noises.  Disconnected the entire AC/DC board and the AC/AC transformer still renders the crackle/crackle/crackle noise during the 0.5s of of magnetisation.  How can a simple grounding fault kill a transformer, or is it a coincidence?

Arrrgh!

Any suggestions would be much appreciated.

[Sgt_Baker]

Furthermore the because-they-were-cheap 7815/7915 I bunged in to replace the obviously-smoked 7815 are reading -8V and +24V - although the datasheets do intimate in a footnote that neither will regulate properly unless experiencing a 5ma load, which they are presently not.  Pulled the big pre-reg 2200uF caps and, while not 100% healthy, they're not nearly dead.

Scope indicates intermittent power loss on +20AC supply, from which +/-15V derive, although that could be the scope its geriatric self. Yet 5VD and +/- 5VA are fine - the mixer still works.  It still... mixes.  Its faders still remember where they're supposed to be, fly around on demand and its innards respond to basic pan/EQ inputs - all of which would be unviable if the unit as a whole were dead.

[Vicus]

Hi, is the metallic body of the 7815 and 7915 well isolate from the heat sink where they are screwed? (I never opened a 01V so some assumpion may be wrong)
I suggest to remove the 2 regulator and check if the input to the regulator is present and stabile. Check with the multimeter (not the best way I know) if there is a short to ground on the output of the regulator. If it's all ok try to install regulator again but probabily they fail again if the problem is still there.

PS: I have a 01V that behave the same issue like the flickering screen and rundom crash.

[mzacharias]

If you are really getting +24 and -8 volts, either you are connecting to the wrong ground for measurement, or the ground itself is no good. Notice the deviation for both is the same - this indicates a shifted (OR MISSING) ground reference.

[Sgt_Baker]

Hi mzacharias and Vicus,

If you are really getting +24 and -8 volts, either you are connecting to the wrong ground for measurement, or the ground itself is no good. Notice the deviation for both is the same - this indicates a shifted (OR MISSING) ground reference.

The -8V / +22V problem was a complete POBCAM on my end.  Embarrassingly so.  The less said the better .  Anyhow, now I have healthy, noiseless, ripple and smoke-free +/- 15V supplies once again it transpires that whatever occurred with the blown -15V regulator took with it the controller-less display's row(?) drivers AND the entire contrast circuit which, irritatingly, is mounted separately on the disassemble-everything-board-of-least-accessibility.  Not that it's relevant.  One can purchase "new" (funky white on blue, don't you know) displays for an O1V for £150 of flesh, which wouldn't address the contrast circuit's problems.  Used O1Vs start at £150.

However! It transpires that Yamaha's traditional attention to detail has indeed preserved the motherboard's SED1335F display controller as it resides, relative to the display itself, behind a buffer IC, which probably saved the entire system's bacon.  And, probing around, all of the correct signalling (except for contrast voltage) appears to be arriving at the far end of the display-end flatflex.  Thus, the course of this particular project has changed to "Can one use a £20 ARM mini-board to intercept the display signalling and at least send over USB or WiFi?"  Currently the answer is unknown.  However cheaper and more educationally gratifying than purchasing a £150 screen where it is already known further work on the contrast circuit is required.

Hi, is the metallic body of the 7815 and 7915 well isolate from the heat sink where they are screwed? (I never opened a 01V so some assumpion may be wrong)
I suggest to remove the 2 regulator and check if the input to the regulator is present and stabile. Check with the multimeter (not the best way I know) if there is a short to ground on the output of the regulator. If it's all ok try to install regulator again but probabily they fail again if the problem is still there.

PS: I have a 01V that behave the same issue like the flickering screen and rundom crash.

The regulators and rectifiers installed are all of the fully-plastic variety.  However the stability problems with your O1V will likely be resolved by installing the ground plane mod pictured in my first post.  Sooner rather than later, as I suspect the -15V L7915 going kaboom may have been a result of my leaving it too long.

[Sgt_Baker]

320 x 80 pixel display. All measurements approximate as the £10 logic analyser clone yet to arrive.

The display's signalling appears to be as follows:

4 x data bits for 16-level greyscale.

Frame-reset clock running at either 5fps or 10fps

Column_CLK running at either 512KHz or 1MHz with a 50% blanking interval.

Row_CLK at 1/2 x 1/80th of Column_CLK.

All seems not entirely beyond the capabilities of an efficiently-coded teeeeny-computer.

[Sgt_Baker]

Why bother, I hear you ask.

Well, it's not only the O1V.  A whole bunch of Special Musical Equipment™ from both Yamaha and Roland used this particular display, now universally failing, and research indicates a plurality of people deplored by £150 replacement displays from eBay which may or may not actually work.  The (Toshiba? Epson?  Hitachi?) communication protocol is universal to all displays.  Thus, with the cost of £20 for the overpowered ARM dev board, and the headache of tacking on eight wires, at least people could once again have access to their XYZ mixer/synth/tone generator's displays so as to navigate menus and save their presets.  And then desolder the board and post it to the next person.  Or keep it.  Or whatever.

[Audiorepair]

Sgt Baker,

do you know whether this issue was fixed in the 01v96i versions?

Or does this problem remain?

[Sgt_Baker]

Sgt Baker,

do you know whether this issue was fixed in the 01v96i versions?

Or does this problem remain?

Hi Audiorepair,

The dodgy grounding issue was reputedly fixed in later revisions of the O1V, mine being S/N #001234, so completely non-existent in the O1V96 as far as I'm aware.  Indeed a quick glance at the 96i's service manual reveals that almost every internal component bears little relation to its older sibling.

[drussell]

Needing to replace the displays in gear like this is why a project like the "universal" LCD controller idea that we were talking about back in places like the 128-x-64-mystery-lcd-identify thread would be a useful device!

[Sgt_Baker]

Needing to replace the displays in gear like this is why a project like the "universal" LCD controller idea that we were talking about back in places like the 128-x-64-mystery-lcd-identify thread would be a useful device!

Yep.  Couldn't agree more.  And the latest experimental evidence is that an Arduino Uno (all I have lying around) isn't nearly capable enough!  Who would have thunk it, eh?

The strategy I'm adopting on this end, owing to the display controller being integrated on to the O1V's motherboard, and thus driving a relatively dumb yet accessible display, is to attempt to capture the post-controller signalling.  Even with the Uno, the results aren't entirely disappointing:


For a 320 x 80 display

Frame changed state to 1 in 14268 uS
Row_CLK_Count/Col_CLK_Count: 79 553

Frame changed state to 0 in 14276 uS
Row_CLK_Count/Col_CLK_Count: 79 554

Frame changed state to 1 in 14272 uS
Row_CLK_Count/Col_CLK_Count: 79 554

Frame changed state to 0 in 14272 uS
Row_CLK_Count/Col_CLK_Count: 79 554


The Uno, with its two hardware interrupts is doing three things:

1) On Row_CLK falling edge increases a counter (Signal LP @ ~5.6KHz in the controllers datasheet, page 17).  79 rows on account of timing issue (1-indexd) or perhaps the controller never writes to the last row.

2) On Col_CLK falling edge increases a couple of counters and writes the data values  (Signal XSCL @ ~1.792MHz x 4bits data = 7.168Mbps) to rotating SRAM.  We can see that the Uno's maximum interrupt servicing rate is pegged firmly at ~38KHz - far shy of the ~1.7MHz required to effectively capture every pixel.  It doesn't matter if one attempts to calculate Pi to a million digits in the ISR.  38KHz is all the little brain can perceive externally.

3) Looping to detect the rising/falling edge of Frame_CLK (Signal WF @ 2 x 35Hz) and writing data out via serial.


Conclusions / Suspicions / Observations:

Janky tacked-on setup possibly ruining the higher frequency signals.

Perhaps looping on the pixel clock and utilising interrupts for slower tasks would be more fruitful?  Potentially slashing the output framerate to allow for slower and more reliable external communication.

Perhaps one of the latest 32-bit ARM-based nanoboards clocked at 120MHz can keep up with the pixel clock rate via ISR???

One of the more expensive FPGA-included nanoboards could almost certainly (?) be completely effective (?) as a Universal LCD signal recorder/rebroadcaster???  To a reasonable point, obviously.  Don't expect to be reinventing the HDMI capture card.

P.S. Was never expecting success with an Uno, just in case anybody wondered.  It doesn't possess enough RAM, at 2048 bytes, to store a single frame of 320 x 80 even if it were squished down to 1bpp.  Just getting a feel for that required to service this particular bee in one's bonnet.

[dubsr6]

Hi Sgt_Baker,

I too am having issues with my old O1v, (S/N 002270).  After experiencing the random turn offs, and the fix (smacking it) every once in awhile. It finally gave up on me and is now suck at the software loader screen, Where it says CURRENT VERSION<<NO DATA>>

I tried hard resetting, but that does not seem to do anything.  Would you know of a next step to try?