SA-3A (LA-3A Clone) Support Thread

[ruckus328]

Alberto, I'm waiting to hear back form a few forum members regarding this matter.  I have some suspicions on what's causing the issue (and a remedy) but would like thier opinions as well before we proceed.  It might take a few days but rest assured we'll get you taken care of.

 

[SSLtech]

Alberto and Mike,

Here's some info from my long hours spent pondering, reverse engineering, dismantling, reassembling, building and cloning T4's:

The opto's don't always have a PERFECTLY linear correlation (so don't expect ABSOLUTE tracking, but within a dB is usually perfectly achievable), and sometimes they age quickly, sometimes errors happen during the matching process etc.

Anyhow,

The LA-3a is GENERALLY similar to the LA-2a in terms of the general design, -as suggested by lots of small details such as the 1k~2k resistor in the 'limit' switch position, the 68k series resistor at the input, the use of the T4 for example-  but two areas where they differ are the availability of VERY high DC voltages for the EL panel drive and for the 'bridge' circuit which drives the GR display.

The EL panel is overcome by use of a step-up transformer to achieve similarly high EL panel AC voltages from a lower voltage DC-fed amplifier circuit... that takes care of that one... but the bridge circuit for the GR meter is addressed by a redesign.

In the LA-2a, tracking is adjusted by adjusting R25. The (large) DC voltage is fed in at the top of the bridge at the wiper of a trim potentiometer, which 'distributes' the current along each differently leg as the wiper is moved to one end and away from the other, and vice-versa.

In the LA-3a, the tracking will have to be adjusted some other way.

Of course the REAL fix is to get a better-matched T4 module, assuming all of the resistances are correct, but if you have an approx. 100k trim pot, I'd suggest temporarily replacing R36 with one, and seeing if it can be made to alter the GR tracking relationship... or if you don't have a 100k trimmer handy, try a 50k trimmer plus 47k in  series, or something similar.... Or try swapping a few different fixed resistances in the 47k to 100k range,  BE ADVISED however, that before swapping fixed resistors, you should wind DOWN the meter ZERO adjustment to something lower, for example about -6dB. This is because the two controls interact massively, just as they will in an LA-2a, but whereas with a trim pot you can watch the interaction in real time and adjust each one to prevent 'burning' the meter by pegging it hard against the endstop, with a fixed resistor swap, a sudden change in resistor value might result in the meter pegging unnecessarily. -You'll probably not REALLY hurt anything, but I'll advise caution here, since this is the first time for trying this... and I don't have an LA-3a here to play with right now.

Hope that makes some sense... I'll take a longer look at the schematic and see if there's anything else obvious, but so far as I can see, the interaction of the value in R36 (plus the meter's series resistance etc) and the resistively-based potential attenuation before it (R22/23 in parallel and the opto-to-ground) should have some tracking effect.

Hope that helps! -let me know what you find, but for now don't go much lower than 47k in position R13, -you might find that going lower makes the problem worse, I haven't thought it through. -Going HIGHER in value might be called for... For now, just try it quickly and see.

Keith

 

[SSLtech]

Here's that part of the schematic crudely re-sketched:

 

[ruckus328]

Keith, thank you very much for your in-depth input.  Your analysis coincides with my suspicion that adjusting R13 (68K) should remedy the situation (in theory anyways).  I was concerned with killing the meter as well with just throwing some random value in there which is why I wanted some second opinions on this before recommending it, good call on backing down the G/R before jumping into it though, I probably wouldn't have thought to take that precaution.

Alberto, real quick and crude way to test this is strap anywhere from a 200K to 250K resistor across R13 (equivalent resistance around 50K).  You'll be able to do it without having to disassemble anything.  Doubt it'll get you where you need to be but it will tell you if you're going in the right direction with lowering the resistance, or if you need to increase it.

 

[SSLtech]

You can (hopefully) see how with the opto completely dark (and thus hopefully completely open-circuit and removed from having any influence) the voltage across the VU meter (and its attendant 3.6k resistor, which is the same resistor which accompanies it EVERYWHERE, even when it's strapped directly across the output terminals to read the AC signal voltage there... consider the two components "as one") is defined by the series resistance above it, which needs to be a certain number... let's say for a simple example to keep the numbers even it has to total 150k ohms.

So... there's a 68k resistor there already; we need to adjust the trim pot to give us another 82k, so -added to the 68k- we get our magic 150k.

We do that... everything is lovely.

With the opto completely SATURATED, we'd get ZERO volts across the meter, because it's shunting all the current to ground UPSTREAM of the meter's takeoff-feed (current taking the path of least resistance and all that) and it wouldn't matter WHAT any of the other resistances are... so our meters will always agree at zero dB (because we have an adjustment for that!) and also at infinite attenuation (because both will be infinite when there's enough light!) -In the middle though, any variations between the CURVES of the opto's turn-on characteristic will require some trickery.

So, now let's consider that we replace R13 with a LARGER value... say 100k. -Now, we'll have to adjust R23 to give us 50k of resistance instead of the previous 82k setting... because we still need the (example) magic number of 150k total. So we do that, and -once again- when the trim pot is adjusted, we have 0dBVU displayed when the opto is dark, and minus infinity when the opto is saturated... but IN BETWEEN will have changed...

-Why?

Well, because if we STOP looking at the meter, R13 etc, and pretend that whole branch simply isn't there for a moment, we can see that we just have a chain of resistance from +27V through R22/R23 (which is set at 82k in one example, and 50k in another example) and an opto, then ground. -At the junction node between the two, when the opto is PARTLY turned on, to some interim value... Let's say 50k for illustration/example... then in the two examples where the audio limiting is producing the SAME amount of light, and driving the opto to the SAME resistance level, (for another example the amount which turns the opto to reduce the output level by 5dB...) then the DC voltage at the point in the GR display circuit (which is what the meter and its total series resistance is 'sniffing' in order to display) then the meter display at that same DC point (which will be somewhere between 0VU and minus infinity) will also be shifted upwards or downwards.

So. We bend the curve to get our match.

Now I don't have any LA-3a's here, but I have worked on plenty. -Previously I've encountered them tracking generally within a dB or so when new, known-matched T4's have been installed, and mistrack when known mismatched T4's have been plugged in. That's been the usual fix, and I've never had to play with any of the resistors.

However, with home-made LA-2a's, it's generally understood that home-made T4's don't have the industrial scale of parts supply nor the facilities to match opto elements to any stunning degree, so the fix of tweaking R25 works quite well... in fact it also means that older T4's which have developed a bit of 'mojo' as the nasty chemicals in the T4 age and react over time (which sound GREAT, but which no longer track perfectly for example) can be adjusted and compensated for.

So... Try playing with R13.

If you can't get the range of adjustment, the optos will need to be better matched... though be aware that most LA-3's with OLD T4's that come to my bench don't tend to track very well... Unless the T4 is dead, in which case once I plug in a known-matched T4, they track just perfectly well.

Hope that explanation clarifies more than it confuses!

Keith

 

[AlbertoM]

Hi Keith and Mike! thanks a lot for your quick and helpful information.

Today I've been tracking some voices (8 hours) through the SA-3A and I'm veeery happy with the sound. The G/R tracking is not a major problem knowing the deviation it has.

I'm not going to have time to play with the resistors for a week or so, but as soon as I can, I'll let you know the results.

;D

Thanks again for keeping all this up.

Alberto

 

[SSLtech]

Hey, you're very welcome.

Let us know how it turns out for you. -If it were me, I'd be tempted to put a 100k trim pot in there, and perhaps if you need more resistance range after having tried it on on its own, solder the 68k back in SERIES with it. -If that don't cover it, ...nothing will!

And -as you say- if you like the SOUND, that's all that matters. LA-3a's are feedBACK detetor units, so no matter what the linearity of the opto in the AUDIO path, they always sound good and predictable.

Keith

 

[okgb]

just starting to stuff
Did this get covered ?

the resisters with * are values that may change or ???

 

[dtrax]

just starting to stuff
Did this get covered ?

the resisters with * are values that may change or ???

I've checked the schematic and BoM, and don't see what you're referring to. Can you be more specific?

 

[leitrim_lad]

Hey there, im just finished wiring up my 1st channel and everything seems to be working but im getting a HUGE buzz that kinda comes and goes.. Sometimes i lose the signal as well and then it just comes back.. Im wondering whats the best way to earth the mic ins and outs.. I just brought a jumper wire on each mic input socket and output socket from pin 1 to the pin beside it thats for the case.. Would it be better to run a wire from each pin 1 separately to the point where the earths are all connected?

Also whenever i seem to touch the front casing or any of the switches it seems to make the buzz better when im holding the mic.. when the buzz is really bad however and i touch the front or a switch etc the buzz seems to get worse..

Any help would be greatly appreciated..

Cheers

 

[ruckus328]

This isn't a mic pre.  Put the mic away, plug in a line level device and see if there is actually a grounding problem. 

In answer to your XLR Pin 1's question though - they can be connected directly to the nearest chassis point or to your star ground, I saw no difference in my setup.  YMMV.  I have both input and output xlr pin 1's grounded, depending on your particular studio grounding setup, you may have to disconnect either the inputs or the outputs.

 

[leitrim_lad]

sorry for the ignorance but wat exactly do u mean by a line levelin device? Cheers

 

[ruckus328]

Just the output of your soundcard.

 

[leitrim_lad]

Cheers for that.. I got it working, ended up that the earth wasnt getting round due to the paint to the case im using, had to run another earth wire to the front panel and my noise was gone.. Still getting a bit of a buzz though, ive recorded in just for you to hear and if you'd have any ideas.. This is with the gain up full  in classic mode and 50db with no peak reduction etc.. iv been moving round the transf but it doesnt seem to make a difference, when i move the lead going to my power switch it helps it a bit but not much.. I noticed when i put my hand down towards the direction of the t4b it gets far louder.. Thanks in advance and sorry about all the stupid questions..

http://www.thevibes.ie/annagh/noise.mp3

Cheers

 

[ruckus328]

Cheers for that.. I got it working, ended up that the earth wasnt getting round due to the paint to the case im using, had to run another earth wire to the front panel and my noise was gone.. Still getting a bit of a buzz though, ive recorded in just for you to hear and if you'd have any ideas.. This is with the gain up full  in classic mode and 50db with no peak reduction etc.. iv been moving round the transf but it doesnt seem to make a difference, when i move the lead going to my power switch it helps it a bit but not much.. I noticed when i put my hand down towards the direction of the t4b it gets far louder.. Thanks in advance and sorry about all the stupid questions..

http://www.thevibes.ie/annagh/noise.mp3

Cheers

The LA3A is not exactly a quiet compressor, although it has been confirmed by many now that my version is quieter than the original.  That being said though, in classic mode with gain at max (50db) I would expect there would be quite a bit of audible noise (gain knob will crank up the noise floor as well as the audio).  Mod & 50db are the recommended default settings, and it should always be in the 50db setting when you're in mod mode, otherwise it will explode and your entire studio will burn to the ground  ;D  Just kidding, it just won't sound very good is all.

The mod mode (also known as the clearwater mod) was a mod done to lower the noise floor, all I did was implement it into my version, it's been around for decades.  The reissue units all have this mode as well.  I almost just ommited the "classic" mode from the design and just hardwired it into mod, but decided to keep it just in case anyone wanted a honest to goodness ol LA3A sound (extra noise and all).

One other thing though, have you ran through the calibration procedure or read through the thread?  There's lots of pointers throughout regarding getting noise to its lowest.

 

[jordan s]

Finally wired mine up and I'm having some problems. When I first powered up, all the lights came on and it looked fine but I almost immediately heard something pop on the second board so I shut it off quickly. It smelled like something had burned up. I checked for any immediate shorts or improper polarization and found nothing so I disconnected the power cable between boards and started measuring voltages across the main board. All the voltages seemed to be fine (btw, what should +V, TP7 read?)
After only a minute or so of being on, I noticed Q5 getting very hot and then R19 started smoking. Of course, I quickly turned it off again. I can't find any shorts, polarizion of all components looks right, voltages read fine. Any clues? I really hope I didn't completely burn something up.

 

[ruckus328]

Jordan, some thoughts:

1) V+ (if you're using transformer 30V toroids with my kits) should read around 40-42VDC.

2) Check Q5 & Q6 are correct components and not reversed (would be my first guess). 

Should be:

Q5 - 2N3053
Q6 - 2N4037

3) Heatsinks installed on Q5 & Q6?

4) Are metal can transistors (Q5, Q6, Q9, and Q11) raised off the board a little?  Otherwise the transistor bodies will short to pads underneath if they are touching the board.

5) If answer is "yes" to all of these questions then set multimeter to measure DC and probe TP2 (WITH UNIT OFF).  Quickly turn unit on and as soon as you get the DC measurement at TP2 turn the unit off as quick as you can as to minimaze any further damage.  Let me know what the DC reading is.

 

[jordan s]

Thanks for replying so quickly. I was really bummed when I stayed up all night to finish wiring this and it started smoking right off the bat.

Well, you certainly called it. I don't know how I screwed it up on BOTH boards but I made Q5 and Q6 2N3053 and Q9 and Q11 2N 4037. Now I just have to figure out what I blew and what I should try to reuse. Oy vey.

 

[ruckus328]

Jordan, I would take the safe bet and replace all 4.  At the very least the ones that were in the incorrect spot, 99% guarantee they are blown.  Inspect the current limiting resistors (you already indicated one is toast).  For the 5 cents each they will cost you I'd replace these as well.  Chances are good nothing else damaged.  Mouser and newark carry these items. 

If you want to cover yourself in case something else got fried, order a few of each type of transistor as well, they don't cost much and if you discover next week one of the other ones were damaged you'll be waiting until the following week for a second shipment for $3 worth of parts, better to just get them all in one shot.

 

[jordan s]

Good call. Looks like this guy is going to be on my bench for another week.