Altec 448A?
- Thread starter eskimo
- Start date
Help Support GroupDIY Audio Forum:
I never have seen that Altec unit.
Seems pretty interesting
Seems pretty interesting
The 444B is limited bandwidth for telephone use. Down 10dB at 10k. Might be a fun "effect" though.
Bo Deadly
Well-known member
Assuming it's the same circuit (it has all the same tubes and ins and outs so it looks like it's going to be close) it might be possible to "fix" the bandpass. I would just measure the response first to see what it is regardless.
But I would leave it. 2W is a perfectly fine practice amp level for guitar. I would put an old alnico speaker on the 4 ohm output and then mic it. It looks like compression is far too slow to hear the compression but it might still have a sound to it or be good for recording clean guitar rhythm passages with very different playing styles / intensities.
I haven't been able to find any info on the peerless and what their frequency response is like.
Bo Deadly
Well-known member
Wouldn't matter if you did. You should measure it anyway. Get a USB audio interface and some software like Room EQ Wizard (REW) and get an FFT spectrum.
I believe it would be wortwhile investigating the frequency response when the capacitors (C3 & 4) across the output xfmr's primary are removed.
Since there is no global NFB, there is no risk of unstability/oscillation).
I don't believe Altec designed/commissioned xfmrs with deliberately poor performance; I'd rather think they used standard parts and srtthe frequency response with said caps.
Again, it's only speculation. My crystal ball is at the voodoo shop.
Since there is no global NFB, there is no risk of unstability/oscillation).
I don't believe Altec designed/commissioned xfmrs with deliberately poor performance; I'd rather think they used standard parts and srtthe frequency response with said caps.
Again, it's only speculation. My crystal ball is at the voodoo shop.
Last edited:
Certainly for low frequency performance. The upper cutoff won’t be affected much here though, would it? Also, I think we can mostly rule out Miller effect blocking the highs with the relatively low gain and 27k resistors of the input stage. Makes me wonder if Altec didn’t save costs by wiring the transformer(s) for the intended frequency response. As I’ve posted here before, I’m curious about this circuit and keep threatening to build one with transformers of choice in hopes of extending frequency response. Really curious about what you find out here.
The only way to reduce the intrinsic HF response of a transformer is to deliberately reduce significantly the coupling coefficient, which iis not easier to do than adding those two capacitors.
I would say that C3 and C4 are almost certainly a deliberate HF bandwidth roll-off. Their reactance at 10 kHz is only 700 Ω, which would produce the -10 dB roll-off if the impedance of the output transformer's primary halves is about 2.5 k (a very reasonable assumption).
Incidentally, the value of the secondary load for any input transformer (the 27 kΩ pair in this case) will determine its HF bandwidth. Set it too high and you'll get a rising HF response (along with a big, hopefully ultrasonic, resonance peak). Set it too low and you'll get a HF roll-off (the load R forms a LPF with the transformer's leakage inductance). There is always an optimum value for this R, usually found on the manufacturer's data sheet (at least for every Jensen) that results in optimally flat frequency response.
Incidentally, the value of the secondary load for any input transformer (the 27 kΩ pair in this case) will determine its HF bandwidth. Set it too high and you'll get a rising HF response (along with a big, hopefully ultrasonic, resonance peak). Set it too low and you'll get a HF roll-off (the load R forms a LPF with the transformer's leakage inductance). There is always an optimum value for this R, usually found on the manufacturer's data sheet (at least for every Jensen) that results in optimally flat frequency response.
I’ve always wondered about the last point there, Bill. Many (most) input transformers in early tube designs were unloaded, yet still had full frequency response (often defined as 30Hz - 15kHz, but sometimes full 20-20k). What about the designs of those older input transformers would obviate the need for resistor loading? The grid resistance looks infinite for all practical purposes, does it not? (It’s this fact that has always made me think that the loading resistor isn’t strictly necessary in tube designs).
Many of these completely unloaded transformers will exhibit huge resonant peaks in response. Presumably, they leave the winding "undamped" to take advantage of the lower side of the resonance to tilt up response at 20 kHz. And it is possible, especially with turns ratios >1:10, to use the secondary winding's own DC resistance and the winding's distributed C, as well as the amplifier's input C (remember Miller C for triodes can be significant), as damping - but remember that the DCR of windings is the only noise source (thermal) in a transformer, so it seems to me a bad tradeoff - a parallel damping resistor generally contributes less overall thermal noise.
Don't forget the source impedance also increases damping. Judicious pairing of the leakage inductance and the expected source impedance cannot be neglected.
These mic pres with an "unloaded secondary input transformer", as specified by RCA in their ribbon mic litt, were designed to be paired with ribbon mics, which have a slightly inductive impedance.
