#21 SM Pro Audio TB202 dual tube preamp / channel strip modding & upgrade

<Further developments can be found in parts two, three and four>

I'll admit, this was a bit of a lucky catch, in that it was mistakenly listed in the wrong eBay category, but one i "stalk" for good deals on certain faulty bits of audio gear. It was up for auction, there didn't seem to be much demand for it, and i ended up winning it for a mere 21 euros. About as much as the shipping for it ended up costing, oddly enough.

But still, even for some 40-odd bucks, not a bad little unit. Two channels, a (bare-bones-ish) compressor on each (1.5-10:1 ratio, 1ms/5ms attack, 500ms/1500ms release, at least according to the specs in the manual), as well as 3-band EQ (80Hz shelving / 1.8kHz bell / 8kHz shelving), and there's a tube / valve involved in the circuitry as well. If nothing else, i figured it would do nicely as a bass DI (or even an overdriven preamp, once some internal re-wiring is figured out and done).


Now, since this was listed as faulty, first thing i did was to jump to Google Image Search to see if i could find at least some photos of the insides, if not even some more detailed data (schematic, service manual, etc). The closest i could get was to find some kinda-blurry photos of what seemed to be an older version / revision of this model, right over here. As opposed to the one in the photos, the unit i caught also has a 1/4" jack for a line input, but is missing those two pushbutton switches next to the input XLR on the back. Some more digging (which led to finding this review of the single-channel TB101) showed that in that older revision, those two switches on the back were for the +48V phantom power and the polarity-flip. On this unit, those had been (much more) conveniently relocated to the front panel, with adding said line input jack on the back.

That being said, i figured the internal differences couldn't be too major. A handful of opamps and a 12AX7 dual triode - no big deal, even if all the active circuitry needed replacing / rebuilding. This, as well as its usefulness / flexibility (not to mention the low, low price) motivated me to go after it. The aforementioned review says the opamps in the TB101 are TL072 for the high-impedance instrument input, and NE5532's otherwise. I might swap the '072's for some OPA2132 or OPA2134 (free samples from Texas Instruments, back in the day), but the 5532's are perfectly adequate otherwise.

That's enough introduction and blabbering, onto the actual DIY-ing / teardown / analysis / repair (if indeed required - wouldn't be the first time i bought something as faulty, which turned out to be, in fact, perfectly fine). External first impressions - a really clean unit, barely any signs of use really, the thing isn't even dirty. A few tiny little scrapes on the top lid, but the front panel's flawless. Well, apart from the rack ears, where some hefty-ish screws have left their imprint, but that's pretty standard for rackmount gear ('cause that's what they're made for, aren't they?). The gain / compression / output pots are butter-smooth, and the EQ controls have a chunky detent in the middle.

I "obviously" needed to go all Dave Jones on it, and not turn it on, but "take it apaaaaah't". Thing is, even if i wanted to, i couldn't have turned it on, due to the power supply being... what it was (and that's putting it mildly). But that's the subject of a separate blog post, in order to keep the length more on the "manageable" side of things.

Inside, same story, all nice and clean, no signs of distress, or "skidmarks", or any blown components whatsoever. Each of the two channels has an NE5532, three TL072's and two TL084's. The power supply seems to provide four voltage rails: +/-15V for the op-amps, 48V for the phantom-power (and pooooossibly the tube as well), and a 12V rail. This latter one, considering the cluster of the two relays and the 2-pin connector going to the power switch on the front panel, is most likely only to power the coils of said relays (both likely being DPST or DPDT), and the series heater filaments of the tube. Nice part, though, is that in addition to the component designators, all the passives also have the values marked on the silkscreen. You don't see that too often, nowadays...

The tube is reeeeeally "encouraging" though - the markings on it amount up to a mere "12AX7B CHINA". Real explicit, eh? And i can't say the mounting method inspires a whole lot of confidence, either. I mean, obviously, laying it horizontally is the only way to fit it into a 1U rack case, but the addition of some extra fastening would've been welcome. Then again, i have my doubts this is supposed to be (or was meant to be) in a touring rack, or thrown around and so on.

One nice attention to detail, design-wise, is the inclusion of pF-range caps in the feedback network of, as far as i can tell, pretty much all the op-amp stages that handle any significant gain. That's done to prevent instability that could produce really high-frequency (think MHz-range) oscillation that would end up causing distortion, if not even thermal-induced destruction of said op-amps. Other than that, the circuit board, at least on the top side, could almost be called scarce. A handful of passives, the odd diode, and a whole bunch of wire-jumpers.

Speaking of the silkscreen, kinda seems to me like this design might've been meant to have more than one way (or possibility) of being powered. Either that, or they just designed the main board to be easily cut in half, to make the single-channel version, perhaps. Or they just couldn't be bothered to remove the unused components from the board design (they all seemed to be marked with an "N", and no values printed).

This thing might be on the cheap side, but that doesn't mean it's been designed totally ass-backwards and/or on a shoestring budget. Upon removing one of the two LED/switch daughterboards, i was quite surprised to see a pair of Panasonic AN6884's driving the level meter for each channel. I'm sure the same thing could've been easily achieved with a couple of quad comparators (jellybean LM339) or bog-standard quad op-amps (LM324, anyone?), and an appropriately-sized resistor divider network. But who knows, maybe they got a good deal on these chips, in bulk. On a semi-related note, all the opamps on the main board seem to be legit Texas Instruments ones (as opposed to no-name brands or knock-offs).

Other than that, as you might've already expected, this thing's gonna need a full re-cap (at the very least for peace of mind, if nothing else). And a part of me is starting to be tempted to hack together a small boost-converter, to bring the tube's plate supply closer to the more normal 150-250V these are supplied with (in guitar preamps and whatnot). The most likely source, as well as easiest (current-wise), will be the 48v rail - the two mic channels won't draw more than 30mA combined; the tube itself will only need single-digit currents for each of the two triodes, and the associated rectifier in the power supply is a 1A diode, so there should be plenty of headroom there. A quick "simulation" confirmed my hunch - a 10mA load at 250V (ie. 2.5W or so) would require roughly 50mA or so from 48V. But on the other hand, the 12AX7 datasheet gives some ballpark values of 0.5mA at 100V plate voltage, and 1.2mA at 250v - all in all, next to nothing, power-wise.

Right, i was saying something about a re-cap: the stock junk ones in here are a mix of "Dumicon" (all capitals, in Brush Script, no less), "Stetron", "Jakec" and "OLC"(?). Obviously, all these WILL have to go - i don't trust'em as far as i can throw'em. On the bright side though, there's not THAT many of them, and it's a single-sided board, so the whole process shouldn't be excessively arduous.

In addition to that, i'm veeeeery tempted to swap out the TL072's for either some NJM2068's or some (more) NE5532's. I won't pretend to be able to hear or point out the exact differences, but considering NwAvGuy's measurements, the 072's are somewhat inappropriate for what's supposed to be a microphone preamplifier, i'd say - not in the least from the perspective of noise, if not also distortion and whatnot. I will, though, "play it safe" a tiny bit, and stick with a JFET-input op-amp for the one handling the instrument input.

I know, i know, the tube's probably noisier than the TL072, but on one hand, since i've got the thing "opened up", and on the other hand, since the tube will be far more easily swapped out than any opamps, i might as well take care of that now. That being said, i'll have to see where i can get my hands on some MC33179's or some other low-noise quad op-amps...

Right, that's enough reverse-engineering. I was mostly curious about the input / tube drive / compressor stages. The EQ is most likely a run-of-the-mill Baxandall tone control (for the shelving low / high controls) and a single bell cut/boost. I went with a method i recall seeing in older Japanese AV receivers and TV's (think Yamaha, Panasonic and the like) - on the single-sided boards those employed, they had the component outlines and designators silkscreen on the bottom of the board, precisely for the sake of serviceability.

Now, i know i've complimented the designers for the helpfully detailed silkscreen, as well as the complete absence of ceramic coupling caps in the signal path, but i do also have a considerable chunk of criticism. I couldn't help but notice a distinct lack of power supply bypass capacitors around the op-amps. I know these aren't terribly high-speed ones, since they only need to handle the audio range (up to 20-something kHz, at least in theory), but it's just good engineering practice to include some 100nF caps - either from each supply rail to ground, or even only between the positive and negative one. No worries there though, got plenty of those to spare, so those will get added on the bottom of the board, nice and discretely.

As it turns out, all the two DPDT relays actually do, is to connect the four power rails to the circuitry on the board; the coils are powered from the 12v rail, and controlled by the power switch on the front panel. A sorta-kinda clever arrangement, i suppose.

Anyway, on to the audio circuitry. The balanced inputs go into a two-op-amp differential receiver. The single-ended signal resulting from that goes through the switched jack of the instrument input on the front panel. The switchable pad follows, and then the signal goes into the first TL072, to which the input gain pot is hooked up, giving a 1x-to-21x gain range (0-26dB or so). The output of this goes into one half of the 12AX7, which is wired up as a DC-coupled cathode follower (one per channel). The output of the triode is buffered by the second half of the aforementioned TL072.

The signal then goes into a voltage divider, made up from a 10K resistor, and the LDR (light-dependent resistor), the "control element" in the optical compressor. The following op-amp does a fixed "make-up gain" of sorts, of about 5.5x (~14dB), but its output forks out in two directions - one going off to the EQ stage, and the other into the detector part of the compressor. The first op-amp of the latter is a 1x-to-23x (0-27dB) adjustable gain stage, which is allegedly the "ratio" control. The next op-amp is a precision rectifier, in order to provide a quasi-DC control voltage. The resistors and capacitors that create the (switchable) attack and release times are next, as well as the opamp and NPN transistor (an S8050) driving the LED, which then subsequently illuminates the gain-controlling LDR. And that's as far as i was willing to go.


Oooooh, you damn bastards... Guess what - the holes in the board, for the capacitor legs, are TIIIINY... 0.6mm tops, which makes the 0.8mm legs of some of the caps i wanted to use (100uF/63V Panasonic FR, 3.5mm pitch & 8mm diameter, for the mic inputs and phantom-power filtering) too thick to fit. Yep, the legs of the smaller 6.3mm diameter caps juuuuuust barely fit through. Ah well, nothing a little drilling can't fix, but it's quite fortunate this is a single-sided board (and as such, no plated vias / through-holes to worry about destroying). And on the bright side, there's only a few of that-size caps needed, so i don't need to drill out the holes associated with ALL the caps. Could be worse, i suppose...

Speaking of re-capping, this will be getting a mixture of Panasonic, Rubycon, Nichicon and Elna caps - some new, some "recycled" from my stash of scavenged parts. Yes, i would and i will gladly trust used Japanese caps way more than brand new Chinese or fake / no-name brand ones. Anyway, Panasonics will be FR 100u/63V for the inputs & phantom-power, and some 10u/50v M and NHG for the level-meter boards; Rubycon-wise, some 100u/25V YXF and 47u/35V MHE; Nichicons are a few 47u/25V VX; Elna are a handful of 22u/63V RE3 ones.

As part of the process, i also replaced the caps in the compressor "timing" circuit. Now, electrolytics are anything BUT precision parts (standard tolerance is +/-20% of the nominal value), and i'll trust a film cap over an electrolytic any day of the week, at least as far as durability and electric leakage is concerned. To this end, the 470nF and 4.7uF caps came out, and in went film caps. Since i had no 4.7uF whatsoever, i soldered in a pair of nice compact 3.3uF film caps, one per channel - close enough. The release times in the spec sheet claim 500ms and 1500ms; a wee bit faster than that won't harm anything and the LDR has a certain "inertia" of its own anyway.

On the subject of film caps, i also corrected an... omission of the designers, in that i tacked on a 100nF film cap between the positive and negative supply pins of each of the opamps - for peace of mind, if nothing else. It definitely won't do any harm; quite the contrary, in fact. The aforementioned opamps did indeed get swapped out - the dual ones, anyway. The DI / gain stage / tube buffer TL072's were replaced by a pair of OPA2132's, and the other four by NE5532's (by sheer virtue of the fact that the only NJM2068's i had were SMD). As far as i could figure, nothing in the circuitry around those pointed toward the requirement of JFET-input opamps, so the bipolar-input 5532's should do just fine.

And that about covers all i needed / wanted to take care of, this time around. At a later date i might consider putting together that plate voltage booster, but i'm in no real hurry with that. Yet another project completed!..