> The so-called "rip-off" here was just the idea of 8 mic pres in 1RU for $99.
As usual, I don't know what we are talking about (I don't read every page of the the SamAsh flyers).
8 preamps for $99 is commercially exciting but technically frightening. Those look like darn good preamps considering they retail for ten bucks each (plus $19 for power and case). I've seen far worse.
In fact some of the puzzling details make sense now: they could have got similar performance with a different and simpler layout, but new layout means design cost and time. They probably had an existing hole pattern for a "better" preamp, stamped 8 of them, and left off the expensive bits.
> I don't know why they bothered with Q1 Q2...
Because (I wonder) it was cheaper to use the existing hole-pattern (from an existing better preamp) and stuff cheap transistors than to do a new layout? Heck, if they buy the "good" transistors (which may not be expensive) in large quantity, it is cheaper to use those than to keep a lesser part in inventory.
Also the 20KHz roll-off makes perfect sense: it doesn't suck but it isn't as shimmery as their high-price (and higher profit) preamps, so it doesn't steal sales (as much).
> Input stage Q1,Q2,U2A has a gain of ((8K2+8K2)/((4K7+4K7)||4K7))+1= 6.2
It may be clearer as ((8K2+8K2+((4K7+4K7)||4K7))) / ((4K7+4K7)||4K7))= 6.2 (so you see where the "+1" came from).
> the top half of the pot (at points less than max) will actually have an additional effect on the gain of U2B as well, eh?
Yeah, but I wasn't going to worry about the control taper. The loading may be "right", or they may have let it be "a little wrong", jumpy, again so as not to embarass their $100/channel models.
> Cross-coupled output stages like U1A,U1B suck.
I do not trust those things. On-paper, or in app-notes, "you can show" that it self-compensates for unbalanced loads, even delivering the same level when one output is shorted. But some of these things are unstable in real life with odd loads. Perfect cross-coupling has to be very exact. This one does not look "perfect" to me, and probably nowhere near instability. I just don't see what the 39K resistors do except confuse things.
> "...gain control from -20dB to +40dB"
Then assuming my rough analysis of the output stage and 1.2 gain was right, and there is another dB or so of error from counting on my fingers (or marketing rounding), that is exact.
What is the use of -20dB gain on a box with 1.3V input overload? (I know: marketing.)
> I prefer not to blow up my gear, no matter how cheap.
Actual smoke is quite unlikely. The cap reduces DC gain to unity (actually about 1.5) and minimizes DC offset. With the 4K7+100 resistors and no C100 the DC gain is 6.2. A random pair of transistors from the same factory reel will have offset voltage around 10mV max. So without the cap there could be 62mV offset at U2A output, blocked by the output cap, quite negligible. If they used two utterly different transistors it could be 600mV, still not enough to reduce headroom a whole dB or strain anything. I don't think the cap does any good at all. And 470uFd is not a negligible expense in a $10 preamp. The only excuse I see (beside oversight) is that they build the SAME board for low-fixed-gain AND for vari-gain products, don't decide which it will be until final case assembly, and then either wire up a pot or stuff a 4K7.
Ya know: if you can use it full-up, 40dB gain, the noise may be better than the usual chips. Several dB lower than a pair of 5534. And as cheap as the 5532 has got, a pair of switch transistor may still be cheaper.
Oh, wait. The noise of the first stage is dominated by the 4K7+100. Noise is 5.6 times higher than a 150 ohm resistor, 15 dB Noise Figure! Saying it that way, it is hard to believe it is usable. But it goes to show that the quest for ever-lower noise figures is often futle specsmanship, and quite high NFs can be perfectly usable for many situations.
I may be wrong about the noise of the 2nd stage, U2B. At mid-rotation the high pot impedance reduces the noise gain. So the 2nd stage may not really dominate until you turn down 6 or 12 dB (34 or 28dB total gain). If the input referenced noise stays within a few dB of theoretical all the way to 30 dB gain, that is not bad.
Also the noise figure becomes moot if you use the modern condensers. Their self-noise is 10-20dB above the noise of a 150 ohm resistor, so an input with a Noise Figure as bad as 6dB or more isn't a real problem. (These mikes are not "noisy" because their output level is 10-20dB higher than a passive dynamic mike.) Although, in the type of situations where you would use 8-for-$99 preamps with $69 condensers, you might be better served by a simple unbalanced (but Phantomed) input and one op-amp (plus inverter for balanced out).
If you want a better noise figure, use a 3-way switch to bridge 1K2 or 100 ohms across the 4K7 cross-emitter resistor, and reduce the value of the feedback resistor around U2B to maybe 1/10th the present value. With the input stage working at gain of 16 or 80, 2nd stage noise becomes a non-issue. And first stage noise goes down with that resistor's value. At 100 ohms (plus the existing 100 ohms) the NF will be just over 3dB, which is good enough for most real needs.
At $99, another trick arises: tack 10K resistors to one of the output pins on the first 7 channels, bus the other end together, and rig a switch to feed that bus into the 8th channel. Walla: basic 7-input mike mixer for $99. Got a seminar conference with a dozen people on stage all talking in mikes, and all you got is an 8-in mixer? Buy a couple of these $99 boxes and lash them up as a 15-in mono mixer. Cheap chaos.
As usual, I don't know what we are talking about (I don't read every page of the the SamAsh flyers).
8 preamps for $99 is commercially exciting but technically frightening. Those look like darn good preamps considering they retail for ten bucks each (plus $19 for power and case). I've seen far worse.
In fact some of the puzzling details make sense now: they could have got similar performance with a different and simpler layout, but new layout means design cost and time. They probably had an existing hole pattern for a "better" preamp, stamped 8 of them, and left off the expensive bits.
> I don't know why they bothered with Q1 Q2...
Because (I wonder) it was cheaper to use the existing hole-pattern (from an existing better preamp) and stuff cheap transistors than to do a new layout? Heck, if they buy the "good" transistors (which may not be expensive) in large quantity, it is cheaper to use those than to keep a lesser part in inventory.
Also the 20KHz roll-off makes perfect sense: it doesn't suck but it isn't as shimmery as their high-price (and higher profit) preamps, so it doesn't steal sales (as much).
> Input stage Q1,Q2,U2A has a gain of ((8K2+8K2)/((4K7+4K7)||4K7))+1= 6.2
It may be clearer as ((8K2+8K2+((4K7+4K7)||4K7))) / ((4K7+4K7)||4K7))= 6.2 (so you see where the "+1" came from).
> the top half of the pot (at points less than max) will actually have an additional effect on the gain of U2B as well, eh?
Yeah, but I wasn't going to worry about the control taper. The loading may be "right", or they may have let it be "a little wrong", jumpy, again so as not to embarass their $100/channel models.
> Cross-coupled output stages like U1A,U1B suck.
I do not trust those things. On-paper, or in app-notes, "you can show" that it self-compensates for unbalanced loads, even delivering the same level when one output is shorted. But some of these things are unstable in real life with odd loads. Perfect cross-coupling has to be very exact. This one does not look "perfect" to me, and probably nowhere near instability. I just don't see what the 39K resistors do except confuse things.
> "...gain control from -20dB to +40dB"
Then assuming my rough analysis of the output stage and 1.2 gain was right, and there is another dB or so of error from counting on my fingers (or marketing rounding), that is exact.
What is the use of -20dB gain on a box with 1.3V input overload? (I know: marketing.)
> I prefer not to blow up my gear, no matter how cheap.
Actual smoke is quite unlikely. The cap reduces DC gain to unity (actually about 1.5) and minimizes DC offset. With the 4K7+100 resistors and no C100 the DC gain is 6.2. A random pair of transistors from the same factory reel will have offset voltage around 10mV max. So without the cap there could be 62mV offset at U2A output, blocked by the output cap, quite negligible. If they used two utterly different transistors it could be 600mV, still not enough to reduce headroom a whole dB or strain anything. I don't think the cap does any good at all. And 470uFd is not a negligible expense in a $10 preamp. The only excuse I see (beside oversight) is that they build the SAME board for low-fixed-gain AND for vari-gain products, don't decide which it will be until final case assembly, and then either wire up a pot or stuff a 4K7.
Ya know: if you can use it full-up, 40dB gain, the noise may be better than the usual chips. Several dB lower than a pair of 5534. And as cheap as the 5532 has got, a pair of switch transistor may still be cheaper.
Oh, wait. The noise of the first stage is dominated by the 4K7+100. Noise is 5.6 times higher than a 150 ohm resistor, 15 dB Noise Figure! Saying it that way, it is hard to believe it is usable. But it goes to show that the quest for ever-lower noise figures is often futle specsmanship, and quite high NFs can be perfectly usable for many situations.
I may be wrong about the noise of the 2nd stage, U2B. At mid-rotation the high pot impedance reduces the noise gain. So the 2nd stage may not really dominate until you turn down 6 or 12 dB (34 or 28dB total gain). If the input referenced noise stays within a few dB of theoretical all the way to 30 dB gain, that is not bad.
Also the noise figure becomes moot if you use the modern condensers. Their self-noise is 10-20dB above the noise of a 150 ohm resistor, so an input with a Noise Figure as bad as 6dB or more isn't a real problem. (These mikes are not "noisy" because their output level is 10-20dB higher than a passive dynamic mike.) Although, in the type of situations where you would use 8-for-$99 preamps with $69 condensers, you might be better served by a simple unbalanced (but Phantomed) input and one op-amp (plus inverter for balanced out).
If you want a better noise figure, use a 3-way switch to bridge 1K2 or 100 ohms across the 4K7 cross-emitter resistor, and reduce the value of the feedback resistor around U2B to maybe 1/10th the present value. With the input stage working at gain of 16 or 80, 2nd stage noise becomes a non-issue. And first stage noise goes down with that resistor's value. At 100 ohms (plus the existing 100 ohms) the NF will be just over 3dB, which is good enough for most real needs.
At $99, another trick arises: tack 10K resistors to one of the output pins on the first 7 channels, bus the other end together, and rig a switch to feed that bus into the 8th channel. Walla: basic 7-input mike mixer for $99. Got a seminar conference with a dozen people on stage all talking in mikes, and all you got is an 8-in mixer? Buy a couple of these $99 boxes and lash them up as a 15-in mono mixer. Cheap chaos.
.
