• Updated 2023-07-12: Hello, Guest! Welcome back, and be sure to check out this follow-up post about our outage a week or so ago.

SE/30 Motherboard: UB10 & UB11 SONY Sound Chip Voltages

aeberbach

Well-known member
Good thread JDW. What I hate most about buying capacitors is when you think you have the perfect cap picked and fill the cart only to find that one value you need isn't made in that range, or worse it is out of stock with a 10 week lead time. Since I have just found another SE/30 I'll be buying more solid polymer soon.

(as long as we don't get into the discussion about whether one cap has better sound stage or imaging than another!)

 

JDW

Well-known member
Good thread JDW. What I hate most about buying capacitors is when you think you have the perfect cap picked and fill the cart only to find that one value you need isn't made in that range, or worse it is out of stock with a 10 week lead time...


Thanks.  I've actually been putting together a YouTube video on the very subject of how to properly select capacitors for recapping projects -- general advice not tied to any particular vintage machine.  It was a part of my research for that video which prompted me to start this thread, actually.  

Back around 2005 or 2006 when I did my first SOLID tantalum recap of an SE/30 motherboard, I purchased a cap kit from a fellow 68kMLA member at the time and didn't give much thought at all to the voltages, especially since the 16V rating on the 47uF tantalums was the same as the stock capacitors.  And even though that machine is still running just fine, the fact remains that I now give a more detailed eye of scrutiny to the capacitors, especially Tantalum which can fail spectacularly in a ball of fire when hit by over-voltage.  Unlike the Tantalum haters (believe me, they do exist on a lot of electronics forums and they think you're nuts for using Tantalum on anything), I still feel Tantalum has it's place when properly chosen.  That's really what this thread is about.  The voltage derating is key.  At some point I will swap out 5 of the 10 47uF 16V caps on my SE/30 motherboard for Polymer versions. Whether I choose 16V or 20V will depend on what's available at the time and the case size.  Some polymer tantalum cases are so small there's hardly any metal (or sometimes no metal) tab on each side, making it very difficult to hand-solder.

Best wishes!

 

JDW

Well-known member
Here's the iCircuit document I created, which will allow you folks to play with the simulation yourselves (if you install iCircuit on your Mac, of course):

View attachment SE:30 Audio Circuit.icircuit.zip

The RMS power through output resistor R10 is about 126mW when both channels are set to 400Hz, using the stock 63Ω speaker.  But when I change the speaker to to simulate the MacEffects clear speaker (which uses 8Ω because I don't think 63Ω in that size are made anymore), I see 838mW on R10!  I mention that speaker because I currently have one installed inside one of my clear case SE/30s.

So far, I've not heard reports of that 8Ω clear speaker causing R10 to burn, which makes me curious about the exact waveform output on pin 10 of UB10 & UB11.  In my simulation document, I have the Left and Right waveform generators (where pin 10 should be) set to an amplitude of 6V AC with no DC offset.  I originally had each set to 12V, but both channels add (in terms of Amplitude) and made a 24V output which cannot be correct. @aeberbach reported measuring 12V peaks across capacitors C3 & C4 (which I assume means at pin 10 of UB10 & UB11), but seeing the exact waveform on a scope would better help understand if my current settings are correct.  I think this is important so we can know if an 8Ω replacement speaker might cause R10 to burn at some point.  I suspect it might put serious stress on R10 because that's a big drop in a resistive load, causing a lot more current flow through that resistor.  R10 looks tiny enough to be only 1/8W.

R10.jpg

By the way, I've Edited this post 6 times, trying in vain each time to set the Width of the above pic to 250px.  It changes in the Preview, but when I click Save, it shows as full size.  I take that to be a forum bug.

 
Last edited by a moderator:

ttb

Well-known member
Alright, unfortunately I have bad news. The SE/30 with solid tantalum caps is my non-working unit, so I am unable to get the comparisons. My two functional SE/30s both have polymer tantalum capacitors. I did, however, measure the 1kHz waveform at the speaker jack (with the speaker plugged in). This is what that looks like on max volume (7/7):

scope_1.png

5.81 V peak-to-peak. It is really quite loud, qualitatively. If someone has an SE/30 with solid tantalum capacitors or better yet aluminum electrolytic, they can use the sound file I linked above and repeat this test so we can compare output levels. Your speaker might be a bit different in impedance, but as long as it's a stock SE/30 speaker (unlike @JDW's) it should be similar. 

If I ever get my third board working, I'll repeat the test on it and update this thread with the solid tantalum output level on the same speaker. Not willing to desolder my polymer tantalum caps for this test :)  

 

JDW

Well-known member
5.81 V peak-to-peak. It is really quite loud, qualitatively. If someone has an SE/30 with solid tantalum capacitors or better yet aluminum electrolytic, they can use the sound file I linked above and repeat this test so we can compare output levels. Your speaker might be a bit different in impedance, but as long as it's a stock SE/30 speaker (unlike @JDW's) it should be similar. 

If I ever get my third board working, I'll repeat the test on it and update this thread with the solid tantalum output level on the same speaker. Not willing to desolder my polymer tantalum caps for this test :)  


Thank you again for your continued testing.

Please don't desolder polymer or solid tantalums unless you want to throw them away.  As per manufacturer instructions, once the caps have been soldered onto a board, desoldering them only to later resolver them would put them out of spec.  The manufacturer recommends discarding such capacitors and replacing with new.  This is especially true for polymer capacitors when have a rather heat sensitive electrolytic.  Think what heat does to plastics.

@ttb, when you say "Speaker Jack" I am led to assume you mean "the output of resistor R10 which feeds the onboard speaker? However, in your earlier post you specifically mentioned a 3.5mm plug in the "phone jack."  I must assume it is still the "headphone jack" where you obtained that 5.81Vp-p measurement.  From looking at the SE/30 audio circuit schematic (see my opening post), it appears that when a 3.5mm plug is fitted into the headphone jack, the connections to resistors R1 & R2 are broken, such that UB10 & UB11 drive the headphone jack directly (the Op-Amp would be unused).   

I am still curious what goes on at the INTERNAL Speaker (output of R10), especially because @aeberbach earlier said he measured "12V peaks" on C3, C4 and/or C5 -- which I assume was measured by him at the UB10/UB11 chip side of the capacitor (the Positive side of the cap). If indeed there are "12V peaks" present on the positive side of C3 & C4, then according to my simulation, those peaks (if present on both C3 & C4 at the same time, and if the peaks were in-phase) would be DOUBLED by the Op-Amp to yield a 24Vp-p signal on the output of R10 (at the Positive side of the Speaker) -- which would not be present on a 3.5mm plug fitted into the headphone jack.  24Vp-p across the internal speaker seems highly unlikely, so if @aeberbach could clarify where those peaks were measured, if they were present on the positive side of both C3 & C4, and if they both were in-phase (which can only be determined on a scope), then we could have better understanding of what was meant by those 12V peaks. I am also very curious if those peaks are measured across C5 too.

There are 2 things we're trying to determined from all this testing:

1. Is there 12V present at any time on the positive side of C3, C4 and or C5?  (If so, a 16V solid tantalum capacitor would be outside it's 50% derating).

2. Does the ESR of C3 & C4 impact the Headphone Jack Output only or also the output of R10 which feeds the internal speaker? (My simulation suggests lower ESR would boost the output at R10 very slightly.)

 

ttb

Well-known member
I measured that 5.81 V p-p at the internal speaker header with the speaker connected. The measurement setup was the scope probe on the positive pin on the back side of J11 with the ground clip attached to a rear connector housing. The previous measurement I quoted was at the external headphone jack on the same computer with the same sound file. In that case, there was no load present, though, so they aren't necessarily comparable.

So this test could address your second stated goal if someone else repeats it with different capacitors and we ignore the part-to-part variation in stock speaker impedance.

 

JDW

Well-known member
I measured that 5.81 V p-p at the internal speaker header with the [stock] speaker connected.


Thank you for clarifying.  Now I am more confused about what @aeberbach said about having measured 12V "peaks".

@aeberbach Can you please clarify that you did indeed measure 0-12Vpeak voltage at the pin-10 output of UB10 & also UB11 respectively using your scope?  And if you put both pin-10s on a scope at the same time, were those 12V peaks in-phase (directly beneath each other on the scope)?  If you measured 0-12Vpeak on both pin-10s and if both peaks were in-phase, as per my simulation, the op-amp would double those voltages to be 24v, which seems very wrong.

Also, @ttb, a measurement of only 5.81Vp-p at the output of R10 with the internal 63Ω stock speaker connected implies only a 2.9v signal at each pin-10, assuming both peaks were in-phase and double their respective amplitudes at the output of R10.  Hmmm...

 

ttb

Well-known member
If I have a couple hours this week I will open it back up and probe the rest of the circuit while playing the tone to provide the full picture. I'd imagine there is some difference from the fact that @aeberbach was playing a different sound, but that can't explain all of it.

 

aeberbach

Well-known member
The sound I played is the ascending tones that happen when you press the programmer's switch button. I'm a bit limited in what I can do because I do not have a booting system set up at all at the moment. I took measurements from the positive side of C3 and C4 - sorry but I'm not going to get the chance to measure again until early January, a holiday is happening.

 

JDW

Well-known member
If I have a couple hours this week I will open it back up and probe the rest of the circuit while playing the tone to provide the full picture. I'd imagine there is some difference from the fact that @aeberbach was playing a different sound, but that can't explain all of it.
Thank you very much for your kind willingness to do that extra testing, @ttb.  Please also try pressing the Programmer's Switch so as to perform the exact same test as @aeberbach.  Curious if you see those +12V (0-peak) voltage spikes too.

 

ttb

Well-known member
That should be even easier. To clarify, we're talking about probing C3, C4, and C5 voltages while playing the ascending tones? 

 

maceffects

Well-known member
I’m very glad to see people with much more electrical knowledge than me work to solve problems and various unknowns. Unfortunately, I have little to add, but I did want to say that some useful information in being shared here, so thank you for that. 

 

JDW

Well-known member
If I have a couple hours this week I will open it back up and probe the rest of the circuit while playing the tone to provide the full picture. I'd imagine there is some difference from the fact that @aeberbach was playing a different sound, but that can't explain all of it.


I re-read all of @aeberbach's previous posts.  He said he measured:

• 5v on the positive side of C3, C4 and C5, individually

• 12v peak voltage on C3, C4 (but he didn't mention if it was also on C5)

It is very important that we know if all 3 of those caps see 12v, even if that 12v is merely a transient now and then.  And yes, measuring on the positive side of each of those 3 caps is important, using a scope.

@maceffects, thank you for your kind words and for following our discussion! :)  

 

ttb

Well-known member
Alright, these measurements are from my other SE/30, recapped with the same caps. All are waveforms of the ascending tones played when the interrupt switch is pressed. I changed the system volume to 7/7 before testing. 

Note: These are all with respect to ground. I also measured across the components directly and the signals were substantially similar.

C3 sees a max of 8.56 V:

scope_0.png

C4 sees a max of 8.67 V:

scope_1.png

C5 is a constant 5 V (plus a fair amount of noise):

scope_2.png

C6 is a constant 1 V or so:

scope_3.png

And finally this is the output at the speaker jack (J11):

scope_4.png

 
Last edited by a moderator:

JDW

Well-known member
Thank you for the extensive testing @ttb.  Again, I cannot remove my motherboard like you, but I am able to take scope voltage readings from the J11 speaker connector.  The video below shows exactly how I tested the stock 63Ω speaker and a MacEffects 8Ω speaker.  As I expected, the 8Ω speaker has substantially less voltage across it because the audio circuit simply cannot supply enough current.  Also note that your measurement at J11 appears to be 2.30V.  I assume that is 0-peak.  Even if you double that to get 4.6Vp-p, it's still much less voltage than I am measuring when pressing the Programmer's Switch to get the death chimes. As shown in my video, my highest measurement was 8.24Vp-p (63Ω speaker, solid tantalum recapped MB).  

I also measured a 13.9V transient at J11 upon power-on, followed by the boot chime showing just under 1.5Vp-p.  

All voltages were much, much lower with the speaker.

So while this doesn't give a comparison with your capacitor voltages, unfortunately, it does show that I am getting higher peak-to-peak voltages at J11 than you are for some reason.



 

ttb

Well-known member
Very interesting. Your peak voltage is similar to mine, but the your minimum appears to be much lower, resulting in the large p-p measurement. I believe that may be because you are pressing the interrupt switch repeatedly (e.g. from a sad Mac). The first signal captured by your scope is therefore the single chord of the sad Mac chime rather than just the ascending tones which is what my wave forms are. For my testing I always hit the reset switch, armed the scope, then hit the interrupt switch to only capture the ascending tones @aeberbach spoke of. 
 

My guess is that if you do the test from a non-sad Mac screen you’ll see measurements substantially similar to mine. 

 

JDW

Well-known member
I believe that may be because you are pressing the interrupt switch repeatedly (e.g. from a sad Mac).


Actually, I got the highest peak-to-peak measurement on my very first press of the Programmer's Switch, prior to getting any Sad Mac.  That measurement at J11 (stock 63Ω speaker) was 8.24Vp-p.

 

JDW

Well-known member
I decided to solder some test wires to UB10-pin10 (capacitor C3), UB11-pin10 (C4), Pin-15 (C5, both chips are tied together here), C9 & C10.  Below are my measurements and a short video of the pin-10 testing. Speaker was the MacEffects 8Ω.

C5: 5v constantly, with or without sound, even during the power-on thump, bong, and programmer's switch press.

C3 & C4: 11v with the power-on thump, and 10.2v with programmer's switch press & chimes.

C9: -12.3v

C10: 12.31v

As such, 16V Solid Tantalum capacitors should not be used at C3, C4, C9 or C10.  (16V Polymer Tantalums are fine though.) The other six 47uF caps can be 16V solid tantalum (or Polymer) because they only see 5V.



 
Top