P2P Redd 47 - a few questions

[earthsled]

I'm considering the attached 48V DC supply. The original transformer is listed as 40V AC. Could the circuit be used with a 45V AC input, or would it need modification?

Parts list:

C1-C4 = 220 uF 100 V
C5,C6 = 10 uF 100 V
D1-D6 = 1N4002
R1 = 82 OHM 1/4 W
R2 = 1 K OHM TRIMMER
R3 = 2.7K OHM 1 W
T1 = 40 V at 0.15 A (6 VA)
U1 = TL783C

Thanks!

 

[abbey road d enfer]

I'm considering the attached 48V DC supply. The original transformer is listed as 40V AC. Could the circuit be used with a 45V AC input, or would it need modification?

Parts list:

C1-C4 = 220 uF 100 V
C5,C6 = 10 uF 100 V
D1-D6 = 1N4002
R1 = 82 OHM 1/4 W
R2 = 1 K OHM TRIMMER
R3 = 2.7K OHM 1 W
T1 = 40 V at 0.15 A (6 VA)
U1 = TL783C

Thanks!

No modification needed for the elevated voltage, the TL783 is rated at 125V and the caps at 100V, and will see only about 60V. just make sure the heartsink is properly dimensioned for the increased dissipation.

 

[earthsled]

No modification needed for the elevated voltage, the TL783 is rated at 125V and the caps at 100V, and will see only about 60V. just make sure the heartsink is properly dimensioned for the increased dissipation.

Is there an equation for calculating the heatsink rating?

Also, I'd like to add a simple unbalanced high-Z input to the circuit. For this, can I tie the signal directly to the grid of V1? (Like what is shown in the attached schematic)?

Thanks!

 

[abbey road d enfer]

Is there an equation for calculating the heatsink rating?

http://www.daycounter.com/Calculators/Heat-Sink-Temperature-Calculator.phtml

Also, I'd like to add a simple unbalanced high-Z input to the circuit. For this, can I tie the signal directly to the grid of V1? (Like what is shown in the attached schematic)?

Yes. You will need to add a high value resistor from grid to gnd. Typically 1Meg.

 

[earthsled]

Again I would suggest a U attenuator, because we don't need to maintain constant impedance.

What sort of values would you recommend if I were to build a "U" style output attenuator on a rotary switch. Would the series resistance remain constant while the shunt resistance is switched?

Also, what would be the effect of using a 5:1 or an 8:1 transformer on the output (as opposed the the original 7:1)?

Thanks!

 

[abbey road d enfer]

Again I would suggest a U attenuator, because we don't need to maintain constant impedance.

What sort of values would you recommend if I were to build a "U" style output attenuator on a rotary switch.

Both questions are related to the load you want to connect it to.
If you connect it to typical 10k line inputs, you have much more latitude than if you want to drive 600r.

Would the series resistance remain constant while the shunt resistance is switched?

That's not ideal.
http://www.quadesl.com/attenuator.html
That's the PI model; you must mirror it to make it balanced. You need to enter half the total value.

Also, what would be the effect of using a 5:1 or an 8:1 transformer on the output (as opposed the the original 7:1)?

The higher the ratio, the lower the output level and the lower the output impedance. Lowering the ratio gives more output level but the output stage may struggle to drive low impedance loads.

 

[earthsled]

http://www.quadesl.com/attenuator.html
That's the PI model; you must mirror it to make it balanced. You need to enter half the total value.

I'm assuming 1K total resistance is needed (500R for balanced) -- is this correct?

Is a shunt resistance needed in addition to the series legs?

Also, is there a way to calculate the steps for equal steps (3dB per step for example)?

Thanks for the link and the info!

 

[abbey road d enfer]

http://www.quadesl.com/attenuator.html
That's the PI model; you must mirror it to make it balanced. You need to enter half the total value.

I'm assuming 1K total resistance is needed (500R for balanced) -- is this correct?

Yes.

Is a shunt resistance needed in addition to the series legs?

The L attenuator is like a potentiometer, so you have indeed a series and a shunt branch.

Also, is there a way to calculate the steps for equal steps (3dB per step for example)?

That's what the calculator offers, equal steps; just define the number of steps and max attenuation.

 

[earthsled]

The L attenuator is like a potentiometer, so you have indeed a series and a shunt branch.

The unbalanced configuration is grounded. For balanced, would I connect the two strings together instead of grounding?

 

[abbey road d enfer]

Exact!

 

[earthsled]

Is there an equation for calculating the heatsink rating?

http://www.daycounter.com/Calculators/Heat-Sink-Temperature-Calculator.phtml

Is this how to calculate the total power dissipated by the regulator?

Pdis ≈ ( Vin - Vout ) x Iout

Vin = 45V AC x 1.414 ≈ 64V DC
Vout = 48V DC
Iout = 14mA x 2 = 28mA (for two channels)

Pdis ≈ 0.44W


Thanks!

 

[abbey road d enfer]

Is there an equation for calculating the heatsink rating?

http://www.daycounter.com/Calculators/Heat-Sink-Temperature-Calculator.phtml

Is this how to calculate the total power dissipated by the regulator?

Pdis ≈ ( Vin - Vout ) x Iout

Vin = 45V AC x 1.414 ≈ 64V DC
Vout = 48V DC
Iout = 14mA x 2 = 28mA (for two channels)

Pdis ≈ 0.44W


Thanks!

Not exactly. You have to add the current in the voltage divider. The 82r res across the 1.2v reference establishes this current at 15mA. That makes a total of 43mA, resulting in a dissipation of ≈ 0.7W

 

[earthsled]

Thanks for your help with the 48V regulator. It looks like it may not need a heatsink at all, but I'll probably use a small one just to be safe.

I have a couple questions about the "rumble filter"... The original EMI schematic shows a 30Hz filter on the input primary. The manual remarks that the filter is "imperfectly terminated" and that it may cause "iron distortion" at certain frequencies. Is there a better location for the low-cut filter in the circuit? Could the filter be modified to have a cutoff between 80 and 100Hz?

Also, I'm still wondering if R2 should be 1/2W or 2W. Using Ohm's law, 205V across a 100K resistor results in 2mA current. 2mA times 205V is about 0.41W. So a 1/2W resistor seems correct for the minimum. Will this be safe, or will it need a higher rating?

 

[abbey road d enfer]

The original EMI schematic shows a 30Hz filter on the input primary. The manual remarks that the filter is "imperfectly terminated" and that it may cause "iron distortion" at certain frequencies.

That's true. The xfmr primarysees an elevated source impedance, which increases the distortionright about the cut-off frequency.

Is there a better location for the low-cut filter in the circuit?

The reasoning behind putting the HPF right before the xfmr is that potential overload due to VLF is eliminated from the start, but it doesn't work to perfection. It is possible to insert the HPF later in the signal path, knowing that any distortion taking place before won't be eliminated. That's the dilemma, and one of the reasons HPF on the mic itself exist. IMO, locating an HPF in the preamp has its justification in giving the possibility of higher order and variability. My take on this subject is building a 2nd-order or higher variable HPF built around a dedicated stage, typically a Sallen & Key structure.

Could the filter be modified to have a cutoff between 80 and 100Hz?

The simple HPF constituted by a series cap (with or without a damping res) can be tuned higher by decreasing the value of the cap. You must experiment to adjust the proper value. The problem ther is that this type of filter can be anything from first to second order, depending on the actual inductive and resistive parts of the impedance. If the input behaves like a pure resistor, the frequency is inversely proportional to capacitance, if it was purely inductive, the frequency would be inversely proportional to the square -root of the capacitance. Since it is actually a mix pf resistor and inductor, the relationship is somewhat unpredictable.

Also, I'm still wondering if R2 should be 1/2W or 2W. Using Ohm's law, 205V across a 100K resistor results in 2mA current. 2mA times 205V is about 0.41W. So a 1/2W resistor seems correct for the minimum. Will this be safe, or will it need a higher rating?

I would use a 1W res there, just in case anything gets out of spec.

 

[CJ]

John Hinson gave me one of his custom made input transformers,

i stuck it an api preamp,

poor guy, what was his handle again? got attacked by abbey road thugs,

Winston O Boogie!  thats it,

he had cinemag lower the ratio from 1:7 , not by much, maybe 1:6 or 1:5.5,

i guess this was to attenuate without resistors,

i bet cinemag would wind a few if you asked them,

 

[earthsled]

ARdE: It was fairly easy and effective to add a single cap between the preamp and output stages of my 1272 clone for a HPF. I wonder if there is a similar location in the signal path of the REDD.47. I'd like to avoid adding active components if possible. Would switching in a smaller value for C4 create a predictable HPF?

CJ: That's interesting info about John H. and Cinemag. I wonder what the advantage of lowering the input transformer's ratio would be. My plan was to use the stock Cinemag CMMI-7C for the input (1:7) and an Edcor XSM15K/600 (5:1) for the output. Thanks for the heads up about Abbey Road's "thugs". I guess Drip electronics keeps his "four-seven" off the radar by selling a kit rather than an outboard product.

 

[abbey road d enfer]

ARdE: It was fairly easy and effective to add a single cap between the preamp and output stages of my 1272 clone for a HPF. I wonder if there is a similar location in the signal path of the REDD.47. I'd like to avoid adding active components if possible. Would switching in a smaller value for C4 create a predictable HPF?

No, it wouldn't work. C4 being inside the open-loop, NFB would compensate the LF frequency response.
You could insert a cap between the secondary of the input xfmr and the grid - you would need to add a grid-leak resistor; this res would be 200-500k, making the caps relatively small (10-4n) for 100Hz cut-off.
Only problem there is switching low-level Hi-Z signals.
Alternatively, you could insert a cap between the output (junction of C5-R13) and the OT. The problem in that case is the cut-off frequency depends on the actual load impedance, and the response has a peak that must be damped, and also there could be more iron-distortion.

I wonder what the advantage of lowering the input transformer's ratio would be.

The lower the ratio, the lower the leakage inductance and the parasitic capacitance, which in turn increases the resonant frequency. Typically a 1:10 xfmr has a resonance just above 20k, which needs to be damped. Reducing the ratio (everything else being constant, of course) to 1:5 gives 6dB attenuation and the resonance goes up to 40k, making it much easier (even unnecessary) to damp.

 

[lassoharp]

Typically a 1:10 xfmr has a resonance just above 20k, which needs to be damped. 

abbey road,

I've seen several classic tube texts mention that a typical transformer coupled amplifier will show a small rising response starting around 2-3K and extending up to 8-10K.  Is this related to the resonance above 20K that you speak of?  Also,  why does this resonance fall in that position with that particular ratio?

Thanks

 

[abbey road d enfer]

Typically a 1:10 xfmr has a resonance just above 20k, which needs to be damped. 

abbey road,

I've seen several classic tube texts mention that a typical transformer coupled amplifier will show a small rising response starting around 2-3K and extending up to 8-10K.  Is this related to the resonance above 20K that you speak of? 

In the old days, when achieving 10kHz was a feat, this parasitic resonance was used for HF response extension. Any xfmr-coupled system will exhibit resonance(s); today, xfmrs are used for their sound, not as a necessity, so the parameters can be chosen more freely.

Also,  why does this resonance fall in that position with that particular ratio?

It's a combination of size, cost, noise factor. Size is more or less governed by max input level vs. THD; it was found that, within that constraint, in order to achieve 20kHz min HF extension, it was not easy to make the secondary impedance higher than 20k. It is actually quite possible to make a xfmr with let's say a ratio of 1:20 with 80k secondary Z, and a resonance higher than 20kHz, but it would be big and very costly (multi-sanwich), and optimizing the active circuitry would be quite difficult.
Triode has too high Miller capacitance, pentode is noisy, bipolar transistor would have to operate with sub-microamps quiescent I'm not sure it would really work, ultra low-noise JFET's are some kind of an enigma (I have experimented with them, could never get the NF I expected with high-Z sources).

 

[CJ]

the lower the turns ratio, the better the freq response,

so a 600:600 utc like the 56  or 66 is pretty easy to wind,

no interleaving required,