L'Fake - A Cheap- A** Expedient to Replace the Elusive SIT

[Antoinel]

Originally Posted by wrenchone

I'm working on a switching voltage/current regulator to fix a fuse -blowing problem with L'Fake Lite that happens when I snap on AC power. I noticed the problem the afternoon befor the latest Burning Amp, and brought the amp configured as L'Fake rather than L'Fake Lite as a result. L'Fake powers up in the living room with no problems, as it has a current source rather than a paralleled light bulb load.

The solution I mention will rguale from a higher DC supply to a lower and limit the startup current. More details will emerge when I get to complete and test the thing...

wrenchone: I watched yesterday a 4 minute video by TI which addressed limiting to a preset value current charging up an empty super capacitor. Unfortunately, I wasted this E-mail; but I'll search for it.

[Zen Mod]

Originally Posted by wrenchone

I'm working on a switching voltage/current regulator to fix a fuse -blowing problem with L'Fake Lite that happens when I snap on AC power. I noticed the problem the afternoon befor the latest Burning Amp, and brought the amp configured as L'Fake rather than L'Fake Lite as a result. L'Fake powers up in the living room with no problems, as it has a current source rather than a paralleled light bulb load.

The solution I mention will rguale from a higher DC supply to a lower and limit the startup current. More details will emerge when I get to complete and test the thing...

heavy NTC startup , with delayed relay bridging will sort that

so , even in case of blackout , you're good

nothing else than classic soft start , with bunch of NTC resistance instead of fixed resistors

[Antoinel]

Originally Posted by Antoinel

wrenchone: I watched yesterday a 4 minute video by TI which addressed limiting to a preset value current charging up an empty super capacitor. Unfortunately, I wasted this E-mail; but I'll search for it.

wrenchone: I found the video:
Go to www.ti.com
Go to Products
Go to Power Management
Scroll down and to the right of the page. Find " Browse All Videos..."
Scroll down the video list and find the title : Engineer it. How to do precision...

I hope it'll help.

[wrenchone]

I have my own discrete switching solution - no steeenking chips! It addresses the problems both of current during startup and regulating down from a higher voltage. All the toroids I have in house will give me unregulated DC outputs of 55-60V. I have L'Fake Light running on a bench at work today with an outboard regulated supply (puzzling and enchanting my co-workers), and it appears to like a ~45V rail.

I thought of using a thermistor, but I still need to regulate down to a lower bus voltage. The switching solution I'm planning is reasonably simple and elegant, and since it also is current-mode, it inherently bucks out the line frequency ripple.

[wrenchone]

I might also add that I've been doing switching power supplies for a living for >30 years...

[Zen Mod]

[wrenchone]

The switching regulator/current limiter is mostly built up, and I'm thinking of going in to work tomorrow to test it. When I'm more sure of things, I'll post a circuit.

[Antoinel]

Originally Posted by wrenchone

The switching regulator/current limiter is mostly built up, and I'm thinking of going in to work tomorrow to test it. When I'm more sure of things, I'll post a circuit.

wrenchone: your expertise in swiching power supplies may imply value in understanding [and assembling] Class D amps. I look forward to the schematic.

My interest in Schade feedback as it pertains to bjts only, had led me to assemble the subject amp as shown in its attached diagram TCAwithFeedback.pdf. It is interesting to apply loop feedback around a current source amp [or transconductance amp; TCA] because this moves the parent TCA closer to a voltage source amp [VSA], and thus imbibes it with the added benefits of improved linearity, increased damping factor, etc. relative to the parent.




Schade feedback in solid state systems seems to have the following requirements found in earlier schematics like those of wrenchone and others.

• An power output "FET"
• Operating as a current source; i.e. its power output is taken from its Drain port.
• Inverts the phase of the input signal presented to its Gate
• Loop feedback is applied from the Drain to the Gate of this FET so as to sum this attendant out of phase feedback signal with the incoming in phase input signal.

The attached schematic has all of the above requirements; except for using an all bjt current source amp [TCA] instead of a power FET. Please note that I am not making claims that this working amp mimicks the sound of a triode or a SIT or other. I am just sharing this practical information which I hope is new and interesting.

• I used a transformer like Mr. Pass did in his F6 to manage [in my case] the requirement of Schade feedback. The amp is stable with the indicated signal phase at the transformer windings. The amp will sing if the phase [for example] on the primary winding is reversed from that shown.
• The phase-inverting diy TCA has a gm = 1 A out/V at its input. It output is comprised of opposed collectors of complementary bjts. It is energized by a dual +/-25 VDC power supply, and its ouput stage idles at ~1.5 Amps [fan cooled]. Loading its output with a loudspeaker [as shown] centers this output's DC offset ~0 V or ground. There are no capacitors in series with the signal train.

It sounded great driving a full range loudpeaker; e.g. MCM 55-1310. It was tonally balanced; deep lows and surprisingly 'out reaching" and detailed highs. The sound was less satisfying when this amp energized ADS/L-730 loudspeakers [three-way] geared to mate with an ideal VSA. The sound from the FR drivers had some character of the parent TCA; but the sound of music with and without loop feedback were discernibly different; after adjusting for the loss in volume due to feedback. I preferred the sound with loop feedback. The extent of feedback [Vf in schematic] maybe relevant to the nature of the sound.

I'll characterize the performance of this system in the region 30Hz -150 Hz [the measurements will be reliable], and publish the data.
Brgds.

[Antoinel]

wrenchone: Here are the promised results. Please note that I used a phase inverting TCA with a higher Gm [5A/Vi] than I reported above [1A/Vi]. This action made this application more realistic as this high Gm is like that of power MOSFETs and JFEts. Like before, I am not making any claims that this bjt power amp sounds like a vacuum tube, a SIT or L'Fake. Hopefully; this is only food for thought and a basis for further experimentation by DIYers.



Page 1 of the attached file [TCAwithFeedback2.pdf] shows the schematic which I used to generate the data shown in page 3. Page 2 is blank. The schematic is like the one shown above; except that I made a provision to measure voltages [the data] without and with feedback [switchable at Vf]. All of the voltages at the indicated nodes in the schematic are displayed in page 3. The input signal Vi from the Function Generator was not changed during the switchover. Here are the findings:

• The amp with feedback was stable [no oscillation]
• The plots of Vo versus Frequency without and with feedback showed that the amp with feedback is still a current source. The key indicator is the constant Vf. This was a constant output current [Rf = 1 Ohm] as the impedance of the loudspeaker [load] varied across a wide range.
• Voltage gain, Gm at any frequency can be readily calculated from the data on Page 3. For example Gm = Vf/Ei ~ 5 with and without feedback. Voltage gain with and without feedback in the plateau region of 80 - 150 Hz was ~40; [or Gm x 8 Ohm].
• A 32% loss in voltage gain due to feedback was observed at any frequency.
• The impact on linearity and damping was not readily evident from these data. Listening tests [without and with feedback] will possibly shed light.

The data from the experiment using the original bjt amplifier of Gm = 1A/Vi were also collected. Same behavior as given above. One difference was the loss in its voltage gain at any frequency by ~10%

I'll assemble the second channel and do a listening assessment.
Brgds.

[wrenchone]

I did a little messing around in PSpice this afternoon, starting out with a current source loaded Schade cell using Darlingtons. If there is no buffer, the distortion into a 10k load is about 0.3% overwhelmingly 2nd & 3rd harmonic . For grins, I tried tacking on my usual p-chanel jfet buffer in front, with the results shown here. I was surpised by the drastic reduction in distortion with the buffer in place.



Edit - in practice, I would use parts in the TO-92L package for a little extra dissipation, as I'm running about 25-26ma of current in the output Darlingtons. The ones I have hanging around are the MPSW45 and a Fairchild TO-92L version of the MPSA63. This might make a nifty little lineamp if you don't mind the signal inversion.