Test Equipment Anonymous (TEA) group therapy thread

[Cerebus]

Two things:

Internationalisation note.

Tonight is when the clocks change in the UK and Euroland. So 00:59:59 GMT will be followed one second later by 02:00:00 BST (Other European time zones are available).

Pootling around in the PHEV today

The dealer had left little more than vapour in the tank (one reason to want to start the day with a full battery charge) so we went out to fill the beast with petrol, do some shopping and have a drive around. Did about 20 miles all told, all in horrible Saturday start-stop traffic. Ate about 3.5 kWh of battery and achieved nearly 52mpg (Honest British gallons, not those short measure US ones). Feels huge to drive after the two seater, even though it's only 710 mm longer and 181 mm wider.

[AVGresponding]

only asked for 7.5A, just in case.
So, no fire risk. Fully supervised,

I keep forgetting that people here know what they are doing.


Fires do happen, usually because the socket or the extension lead are not up to the task. A 10A load all night long is different that a kettle running for a few minutes. 







 

On the other hand, if you have a 0.75mm² lead for a 2kW heater, the flex becomes just another heating element...   

[AVGresponding]

"And as the stridulent Discordance rang out, he thought to himself, 'fuck me, I need another drink'"

[Specmaster]

Yesterday morning this is not what I'd have told you was going to be my first job today after making coffee:



Bejezzus this thing is complicated. I spend most of yesterday evening reading the handbook, trying to learn the 1001 settings and controls on the car. What happened to "Stick the key in, turn it, go"?

You do realise that is unsafe, right?


Trip hazard aside, that is a fire risk.    The EVSE lead has a temperature sensor in the plug.  It will reduce the current or even stop the charge if the  plug or socket overheats.

An extension lead does not.

[sarcasm on]
Gosh no, I never considered that reading the rating plate on the back of the charger first (which says 10A) which is explicitly designed for charging from a domestic 13A supply, and plugging it into a 13A rated extension cable that had been fully unwound, with a properly fused plugtop, plugged into a radial circuit with a 16A ELCB wouldn't be OK. Thanks for educating me. I'm also having problems learning how to suck eggs. Could you help?
[sarcasm mode off]

There is no fire risk, at least no more than the 3kW kettle in my kitchen also presents.

There's a world of difference between a one off, let's see if this PHEV charges, to a permanent installation that can be used any time, unsupervised, in any weather. It's a dry day, I was pootling back and forward cleaning windows, exploring features found in the manual, and generally keeping an eye on things, and of course I checked ratings, loads etc. before starting, heck I even put the car in "reduced charge demand" mode beforehand so that it only asked for 7.5A, just in case. I ought not to need to point out the minutiae of all that, here of all places, before someone takes it upon themselves to offer ill-founded advice and implicit criticism. I'd like to think I've demonstrated my technical chops often enough here to avoid that.

So, no fire risk. Fully supervised, so if I saw someone coming who it looked likely that would present a genuine trip hazard to (the old, very young, infirm, or congenitally stupid) I was on hand to say "Carefully now" and "Down with this sort of thing".

Tsk Tsk, C, I don't believe for a single moment anyone here was having a pop at you in any shape or form. I read the comments as being more a generalisation about the overall problems associated with PHEV's. Perhaps this is where the problems occur when things get a bit heated between some members on here, maybe it's because the context has not been fully understood. for instance, the cable on the extension lead is not designed to be used in such fashion where prams, mobility aids and even footfall are all going to be grinding it against what is a pretty abrasive surface (footpath, paving slabs etc) that could put the cable's integrity at risk etc. This is why I mentioned town and city dwellers, very few of whom have a garage or drive that they could locate a proper charging point safely. What about those devils living in high-rise flats, where and how do they charge their PHEV

I've said it before, and I'll say it again, this whole policy of switching over to EV's needs far more thought and some proper action to resolve the practical issues. Take the case of the high-rise dwellers, are they purchase really long extension leads and drop them out from their windows to the ground and some of those poor devils even have to park in another street altogether, how do they get their EV charged overnight  Then of course, that says nothing at all about the physics involved with massive losses in the length of cable (ohms law) etc.

[mnementh]

I can't hear you... too busy nuking a dirt-sucker... again. 

mnem
*pokes the trans-motivator with a stick*

[BU508A]

Hello TEA community,

today I'd like to introduce a Keithley 225 precision current source which I received this week.
It is a current source with seven ranges, starting range from 10nA (with a resolution of 100pA) and going up to 10mA with a max. output current of ca. 100mA.
The 225 has been introduced in the 70's and my device is from 1979, Rev. D (printed on the PCB).
I did some measurements with my DMM 7510 and I'm impressed about the performance of the 225.
The voltage reference being used is a 1N827A  (D116). Manual can be downloaded from here (xdevs.com).

But now for the pictures:

Front, back and a detailed view of the connectors on the back.
Some writing on the back is saying August, 29th, 1979.


Top and bottom side of the PCB. As you can see there is not so much in it.
The gold plating of the switches looks massive.


Some detailed pictures of the electronics, starting with some bogroll capacitors, some OpAmps with date codes, some bodges and a thermocoupled transistor pair


The bodges in detail and the thermocoupled transistor pair


range selector switch with a potentiometer at its end for the fine tuning, another view to the range selector, output selector switch


Just in case you are missing a picture of the reference zener (1N827A),
it is directly mounted on the output selector switch (the red one):


Range selector switches, top and bottom view, mains switch (shielded)


Output connectors, front and back


How does it perform? I did some measurements.
Well, I've measured all seven ranges and did a picture for every dial in that range.
The pictures are starting with the LSB dial.
Setting of the Keithley DMM7510: 10 NPLC, line sync on

Range: 10nA, minimum resolution step: 100pA


Range: 100nA, minimum resolution step: 1nA


Range: 1µA, minimum resolution step: 10nA


Range: 10µA, minimum resolution step: 100nA


Range: 100µA, minimum resolution step: 1µA


Range: 1mA, minimum resolution step: 10µA


Range: 10mA, minimum resolution step: 100µA


Schematic, shamelessly borrowed from this thread:

 

Hope you've enjoyed this teardown and found it helpful.

Edit: typos, added some details
Edit 2: added some detail pictures of the bodges and the reference zener

[dl6lr]

You can still buy Pizzas there though.... but it's a machine / distributor only. Stick your credit card in, press a few buttons on the tactile screen and it gives you a pizza....

Wow, THAT is NASTY...

[factory]

Sort of TEA related for a change, failed Sprague 109D wet tant  capacitor, rectifier diode is shorted too and the acid that escaped from tant ate the track underneath. 





David

[McBryce]

IMHO PHEV are pretty pointless. It's greenwashing in its worst form - a lot of pointless (even harmfull) additional effort for little gain. How much recuperation can this thing do, if even any at all? Because the only two upsides I can see for PHEV in general, are charging the battery while breaking a lot (think driving downhill in the Alps), then using that charge on the next uphill, and not having to turn on the ICE at all just to go buy groceries. One is hampered by the limited charge rate of small batterys, the other by the sparse charge opportunities in cities. While both will get better over time, it still seems like bandaid at best.

It is for most people who need a car. But sadly, before we can get to the point of changing legislation, we have to beat the legislators up with proof of demand for real EVs in the Court of Public Opinion, and we do that most convincingly by voting with our dollars.

It is a maddeningly slow and exhausting process, and mired in bureaucracy like a trying to wade through an ocean of Jello... or maybe a city-sized Roach Motel...    but we cannot just ignore the Band-aid choice until we are able to force them to give us a real choice.

EDIT: Sadly, this dysfunctional model also applies to your comment on household solar above.   

mnem
*toddles off to make a toddy for my body*

Be glad you're just observing it from the outside. This is what I do for a living!

McBryce.

[Specmaster]

Sort of TEA related for a change, failed Sprague 109D wet tant  capacitor, rectifier diode is shorted too and the acid that escaped from tant ate the track underneath. 





David

Thats nasty.

[Vince]

Discord time, EV related traffic is calming down, I can chime in now ! 

Been busy yesterday and today working on the 2467B PSU recap, been skimming TEA.

@Spec : will have a look at that video. Congrats on the miniscope... now that you have cables to split the boards, troubleshooting the second issue with the vertical amp should not take long...

@ TerraOperative : didn't know TI made their own special IS socket, will need to look into that, see how they are made, what's wrong with them...

So the recap was a chore, a roller coaster, did not go smoothly at all. Cap replacements not being 100% identical required adaptation and "oh well it's gonna stay like that no choice", mindset.

Putting new caps in was tedious and time wasting because I could not clean the holes before inserting them, because no fancy power desoldering tool here, only solder wick which did not work as the holes were usually too small.
Soldering did not go well either. My old Weller Magnastat is only, which did not cut it on many occasions. Every time a lead was connected to the ground plane, could barely melt the joint, double-sided at that. Damaged or ripped pads, lifted tracks, followed. Repairs had to be done.

The non-polarized 1uF 50V cap, redial, tiny, was replaced because no better choice, yet still mega expensive, by an axial 1uF 100V not 50V. I thought OK no worries I can just mount it vertically... but no, because somehow the cap turned out to be TEN times bigger than the old one !     Mounting it vertically would have been like erecting the empire state building on that board, just no way, if just for the risk of them interfering with something when I would put the PSU and scope back together. So I had to mount them (2 of them) horizontally...

Then I put the PSU and scope back together. Somehow it went pretty well. I mean I 50% remembered how to put it back together. Then for 30% of it I improvised successfully, and for the remaining 20% I was rescued by the pics I took when I took it apart. Always take pics when you take stuff apart....
At first I bitched at the golden "combs" pins that connect the two board back to back.. how to align them ! 
... then I realized you can remove the combs, there is a plastic clips for that, then you screw the two boards together, then you can just easily insert the combs back into place...

Then I went to replace the couple electrolytic caps on the CRT board.

Power up time... no explosions or smoke.. good ! 
Then I checked voltages, (quick before the ASICs start getting hot), following the service manual. First the 10V reference rail. It's very tight, tolerance only +/- 10mV !  i.e. 0.1%
So that's the 0.1% case where my Fluke 11 daily driver DMM and its 0.9% just won't cut it...
So had an excuse to use a Metrix MX56C instead, at 0.025% it was about good enough for the job. Just.
Cold, right after power up, it measures at 9.995V, which was already well within spec. It was climbing slowly. After 30s or a minute, say, it settled at... 10.000V ! That's one good scope, 30 year old and doesn't even need to touch up the reference rail ! 

Then I checked all the other rails, which were all wellll within spec.

I was happy with myself at this point... and allowed myself to have a look at the CRT to see how the scope was doing. Almost had a heart attack.. I got what you see below : a very, very sick puppy !
Nothing but a super fat blurry flat line, with a few big bright spots embedded in it. 
First thought was to cry and shift it on Ebay for parts to try to get back some of the 300 Euros I paid for it almost 2 years ago.
Then I recovered my senses and tried to see what I could do. Tried diagnosing it.

1) When starting, I got the expected LED and relay dance, so looked like the CPU board was fine.
2) Could not focus the trace, so something wrong for sure on the CRT side of things.
3) Playing with the intensity controls, I noticed that I could make the bright blobs come and go using the readout intensity control.
4) vertical position control is unresponsive.

So... I thought hmm...maybe the problem is 100% CRT related, and the vertical deflection is toast. The bright blobs are probably just the various readout elements on the scree, being squished.

So I looked at the connector in the CRT area and found a lone wire with a Dupont connector at the end, lose, next to that small ceramic board mounted on the CRT neck. Reconnected it.. problem fixed, scope back to life ! Phew that was close !!   


So then I could, for the first time since I bought that scope ages ago, give it a more in depth test ride.

Found some big issues....



1) Vertical amps for Ch1 and Ch2 have huge offsets ! 

I measured them all, see below. At worst, offset makes the trace disappear off the screen ! 

I centered the trace on the 5V / div caliber, then cranked the attenuators and noted the offset.

Calibre (V) | Ch1 (Div)| Ch2 (Div)
-------------------------------------
5           |  OK      |  OK
2           |  -0.1    |  -0.75
1           |  -0.2    |  -2
0.5         |  OK      |  OK
0.2         |  -0.1    |  -0.75
0.1         |  -0.2    |  -2
50m         |  OK      |  OK
20m         |  -0.1    |  -0.75
10m         |  -0.2    |  -2
5m          |  -0.9    |  OFF screen
2m          |  -2.5    |  OFF screen


Looking at the table summarizing everything, there is clearly a pattern here. Apart from the most trow sensitive settings, 2 and 5mV, whic are much worse than the other, all the other settings follow a repeatable pattern / sequence.

Looks like the attenuator is made of two section in series. One that handles the decades, and another that does the 1/2/5 steps thing.
Looks like the decade section is working fine, but the 1/25 is not. Two fo the steps generate an offset, whic get "copy/pasted" as is, in every decade.
Offset is constant no matter the decade.  Ch2 has worse offset levels than Ch1, but does follow the exact same pattern.

Hmmm.... that's suspicious now I think of it ! The problem must therefore lie in some circuitry common to both channels...

I will check the service manual to see if there are adjustments I can make. If not, well I guess I have dead amplifier ASICs, great.



1) Time base problem :

Now for the second problem... I have not spent considerable time looking into this one, but enough time to witness it and confirm it... there is definitely something wrong here...
It happened when I briefly played with the second/delayed time base. I would work the big sweep speed knob, the CPU/readout would change/update the sweep speed accordingly, but... the trace on the screen would NOT budge ! It would not ACTUALLY modify the sweep speed ! .. but if I keep cranking the knob, it eventually does so. Weird.
Moving back to the main time base now, I didn't experience this problem.

sub-problem : once I was in Time base B mode, after a minute, the signal trace suddenly becomes extremely unstable, showing huge amount of horizontal "jitter", like, a few minor divisions. Scary.
Again, problem disappeared as soon as I switched back to the main time base.
So again, maybe another dead ASIC, great.... 



As for the already known, minor issues :

A) dead statu SLED : it's not dead, my recollection was wrong. The problem is simply that it's dimmer than its neighbours. But it does light up. Good news is, it's the status LED for enabling Ch2.
That means, the CPU will need to control this LED individually, hence unlike 5% of the LEDs on the front panel, this one will NOT share it's series limiting resistor with other LEDs. So, I can play with the resistor to adjust brightness when I replace the LED, to try and match it with the other LEDs.


B) The readout on the screen is still very "noisy" and a bit giggling, unstable. however the signal trace is perfectly clean and stable.
So it's not a PSU or CRT problem, I would think. Since the readout is generated on the CPU board, maybe that board needs a recap as well... there are quite a few in there, including dipped tantalums...
However since my soldering iron is way underpowered to handle this kind of board, and I don't have desoldering gun... I will NOT touch this board until I have better gear !!!!
I guess the readout signal path also traverse the main analog board, so maybe the noise and instability are introduced there instead. This board does have electrolytic caps as well.

C) Backup battery. I looked at how much space was available between the CPU board and the scope cover. LOTS to my surprise ! Eyeballing it, a replacement battery can be well over twice the thickness of the OEM " flat " shaped cell, yet still clear the cover. So my 14.5mm  axial leaded " half AA " Varta Battery will fit juuuust fine, zero worries there, good ! 

So that's it... if I do have 2 dead ASIC then I guess the scope is good for scrap anyway, but for now I will leave it at that. Will replace the battery and recap the CPU board in a few months when I hopefully can afford some better soldering and desoldering gear, and take it from there.  Maybe by then I will have collected information about these issues and be able to tackle them... if there is hope.

So to sum it up, recap went well, but scope might still be scrap later this year 

[srb1954]

Edit.....or some dumb ass could come walking by and trip/fall over the cord and sue you.

The worst someone could do to it is unplug the mains plug end, the car side is locked. I've run a mains extension into the street outside this house countless times for all sorts of purposes, some attended and some unattended. Nobody has ever tampered with it, nobody to my knowledge has tripped over it (if they had I'd have probably heard the swearing through the open living room window).

Even if the car side is locked a vandal could do some expensive damage to your car trying to forcefully remove it.

[Robert763]

The problem with the TI low profile sockets is that they made contact with the edges of the IC leg. This has a couple of issues. First is a small contact area. This is intially hidden by the second issue. The edges of the IC legs are rough because of the stamping process. The roughness removed the plating on the socket contacts. Initall the connection is OK because the two parts bite together but eventually corrosion sets in and you get intermittent contact.

[Vince]

Thanks for the detailed explanation. Looked them up in Google Images to see what they actually look like.. then looked at my drawers / stock...

Turns out I have a lot of them ! 

50% of my 14 pin sockets are TI.
90% of my 16 pin sockets are TI.

Oh well.... 

[mansaxel]

On the other hand, if you have a 0.75mm² lead for a 2kW heater, the flex becomes just another heating element...   

That would be stupid, so why do that? I still cringe from 13A loads on 1,5mm² because I was brought up on 10A for that cross section.

[Vince]

2467B update.

Just spent a few minutes playing / investigating the delayed time base issue....

The way it behaves baffled me, but the more I played with it, the more it looked like there was some logic to it...

So much so that I now, fingers crossed, I would say that there is no problem with that delayde time base... I think I now understand how it works, how to operate it.
Now that I understand it, I think it all works just fine, and is quite nice to use.

The thing is, it's my first 24XX scope and the delayed time base is does not wqork the same way as it does on all the 2000 series scopes, what I was used to so far.

so maybe no defective ASIC then... the magnitude of the feeling of relief that invades me right now, is impossible to convey. Let's just I am happy 

Another "problem" I fixed... I always found the trace on this scope sub standard : not sharp. Acceptable, like you would find it on a cheap basic analog scope, but not much better.
Was hoping the PSU recap would fix it, no joy.  So I was starting to give up on it and just say it must be a downside inherent to this MCP CRT, somehow.
Then I had the idea of trying to enable the 20MHz B/XW limiting feature... hey presto, trace looks sharper now ! 
It's my first high B/W analog scope, I didn't realize they could be picking up so much noise that it would affect the thickness of the trace that much.


So now the only major problem is the DC offsets in the vertical amps. If I can fix that too, i.e. hot an ASIC problem, then this scope would be a winner...

Below the rise time of my old Philips sig gen. 200us, using the delayed time base, and hiding the main time base.

[Cerebus]

Edit.....or some dumb ass could come walking by and trip/fall over the cord and sue you.

The worst someone could do to it is unplug the mains plug end, the car side is locked. I've run a mains extension into the street outside this house countless times for all sorts of purposes, some attended and some unattended. Nobody has ever tampered with it, nobody to my knowledge has tripped over it (if they had I'd have probably heard the swearing through the open living room window).

Even if the car side is locked a vandal could do some expensive damage to your car trying to forcefully remove it.

Yes, and a vandal could do massive damage to the whole car with a 4lb club hammer. That's the nature of a vandal. C'mon, this is getting ridiculous. What's your proposal, armed guards for all charging points that aren't behind an electrified fence?

[Cerebus]

On the other hand, if you have a 0.75mm² lead for a 2kW heater, the flex becomes just another heating element...   

That would be stupid, so why do that?

He's probably encountered builders who go "It's OK, I'll do the electrics myself, a fool could do it.". The soto-voce response should be "Yes, and they will be.". I've seen some of the horrors perpetrated by builders of that ilk.

I still cringe from 13A loads on 1,5mm² because I was brought up on 10A for that cross section.

Personally I find the temperature rise that some of the standard cable ratings are based on a bit too much. Yes, they are safe and well within the limits of the materials involved, but I find the idea of an immediately palpable temperature rise a little unnerving. I find myself regularly derating cables or going up a size. I also find myself wanting intermediate CSAs outside of the common standard series, the jump from 1.5mm² to 2.5mm² is a big one mechanically in terms of "handle-ability" when you're reeling or unreeling a bunch of it.

[Specmaster]

The problem with the TI low profile sockets is that they made contact with the edges of the IC leg. This has a couple of issues. First is a small contact area. This is intially hidden by the second issue. The edges of the IC legs are rough because of the stamping process. The roughness removed the plating on the socket contacts. Initall the connection is OK because the two parts bite together but eventually corrosion sets in and you get intermittent contact.

Yes, that is a real pain, I've had that happen to me before, IIRC on one occasion, removing the chip it actually tore a leg off because the roughness allowed some corrosion to set in that was hidden from view.

[Specmaster]

On the other hand, if you have a 0.75mm² lead for a 2kW heater, the flex becomes just another heating element...   

That would be stupid, so why do that? I still cringe from 13A loads on 1,5mm² because I was brought up on 10A for that cross section.

The thing is, there are people out there who are clueless and will do it out of sheer ignorance. I saw a report the other day that claimed domestic extension leads are a mayor cause of house fires. With the increase in gadgets these that most houses have, there is never enough sockets in the walls to support them. Take my own living room, for example, behind the TV there is a twin socket outlet which is not enough, so I have 2 extension leads, into which I have the following plugged in, TV, Stereo, Audio streamer, Sub Woofer, Sound Bar, Free view recorder, Blu-ray player, Ethernet hub and a WAP. It is true that not all things are in use at the same time and it is also very true that the total loading is well within the cable size as the loading in total is only about 2A. But less knowledgable people would happily plug electric heaters, hair dryers etc in and use them at the same time, and in some cases, even cover the leads over, or lay them under carpets and rugs and then walk on crushing the cable and placing even greater strain on their safety.

[med6753]

7904 fan update.

I've been running the scope with the fan connected to the external 6215A supply to get an idea of how well it's cooling the scope. The fan is located on the right side and it seems to be doing a good job in exhausting the heat. But only on the right side. The 7904 has an extensive center "back bone" running front to back. Result is that the vertical board on the left side is not getting much benefit of the air flow. So the top cover on the left side is still very warm. I can see why there's a recommendation to run 2 fans.

I've decided that I am also going to run 2 fans. So I'm going to run a separate power supply for the fans only. I've already salvaged a suitable transformer and a perfect place to mount it. I'll provide an update when it's installed. It will easily connect into the scope's power switch as well as fuse protection. I found a fan in my stash that's approximately half size of the currently installed fan which will fit nicely on the left side.

The salvaged transformer.



Smaller 12V fan.



Unlike my infamous 2465 fan bodge this install is very neat. The only clue something is up is the 4 mounting screws. The fan itself is whisper quiet.

     
 

[mnementh]

You can still buy Pizzas there though.... but it's a machine / distributor only. Stick your credit card in, press a few buttons on the tactile screen and it gives you a pizza....

Wow, THAT is NASTY...

mnem
Wait til you see it. 

[mnementh]

On the other hand, if you have a 0.75mm² lead for a 2kW heater, the flex becomes just another heating element...   

That would be stupid, so why do that? I still cringe from 13A loads on 1,5mm² because I was brought up on 10A for that cross section.

The thing is, there are people out there who are clueless and will do it out of sheer ignorance. I saw a report the other day that claimed domestic extension leads are a mayor cause of house fires. With the increase in gadgets these that most houses have, there is never enough sockets in the walls to support them. Take my own living room, for example, behind the TV there is a twin socket outlet which is not enough, so I have 2 extension leads, into which I have the following plugged in, TV, Stereo, Audio streamer, Sub Woofer, Sound Bar, Free view recorder, Blu-ray player, Ethernet hub and a WAP. It is true that not all things are in use at the same time and it is also very true that the total loading is well within the cable size as the loading in total is only about 2A. But less knowledgable people would happily plug electric heaters, hair dryers etc in and use them at the same time, and in some cases, even cover the leads over, or lay them under carpets and rugs and then walk on crushing the cable and placing even greater strain on their safety.

And lets not forget how they had to change the plugs on Xmas lights to have those tiny little fuses and made them all non-polarized so it's harder (the hardcore dumbasses will just shove until they make it fit or the socket explodes) to plug anything except another string of Xmas lights into them.

Because newspaper and insurance industry archives are full of horror stories about people using Xmas lights as extension cords. 

mnem
The stupid makes my brain cry...

[mnementh]

Smaller 12V fan.   

    BWAHAHAHAHA!

I recognize that fan... I usedta have somewhere around a gross of the fucking things; Dell used 'em in everything desktop for almost 10 years. Have you actually test run the thing in situ...? I ask because I know these are thermally controlled... and once they feel the the heat, they ramp up pretty aggressively. The turbo noise can be quite impressive.

mnem

[0culus]

7904 fan update.

I've been running the scope with the fan connected to the external 6215A supply to get an idea of how well it's cooling the scope. The fan is located on the right side and it seems to be doing a good job in exhausting the heat. But only on the right side. The 7904 has an extensive center "back bone" running front to back. Result is that the vertical board on the left side is not getting much benefit of the air flow. So the top cover on the left side is still very warm. I can see why there's a recommendation to run 2 fans.

I've decided that I am also going to run 2 fans. So I'm going to run a separate power supply for the fans only. I've already salvaged a suitable transformer and a perfect place to mount it. I'll provide an update when it's installed. It will easily connect into the scope's power switch as well as fuse protection. I found a fan in my stash that's approximately half size of the currently installed fan which will fit nicely on the left side.

The salvaged transformer.



Smaller 12V fan.



Unlike my infamous 2465 fan bodge this install is very neat. The only clue something is up is the 4 mounting screws. The fan itself is whisper quiet.

     
 

You should also see if adding the fan causes the time base to drift and the like. The 7904 was clearly designed to be cooled by convection. The 7904A has a factory fan though.