you can find Joe Rasmussens amp here:
https://www.diyaudio.com/community/...tt-transconductance-current-amplifier.239321/
Are you sure this formula is correct?
I did the calculation by using the lowest guaranteed gain of the LM3875 (90dB) and the result (449)is far from the value that was measured by Joe Rasmussen (240...270, depending on frequency)
https://www.diyaudio.com/community/...tt-transconductance-current-amplifier.239321/
Are you sure this formula is correct?
I did the calculation by using the lowest guaranteed gain of the LM3875 (90dB) and the result (449)is far from the value that was measured by Joe Rasmussen (240...270, depending on frequency)
What means musical timing in reproduction with a speaker? My thought was that an instrument (violine for example) has a exactly defined spectrum in the time-domain. Assuming this is correct recorded this timing shows up correct at the output voltage of a voltage amp. But the speaker shifts the phase of the current when voltage driven, so i assume the exactly defined spectrum is also shifted - the overtones showing up a little timeshifted, because of the phaseshift in the current. So in this case, the violine sounds a little different. Probably not totally different, but i hope you get the idea. But that is no longer true if the speaker is current driven, if i understand it right.
And yes, the world thinks in voltage. If it comes to current the world seems to be a totally new one - even for specialists. That includes me too.
Bernd
And yes, the world thinks in voltage. If it comes to current the world seems to be a totally new one - even for specialists. That includes me too.
Bernd
Timeshift is not the same as phase shift. You can have lots of phase shifts, meaning the signal changes shape, but it still gets to the output in nanoseconds.
Its a hard thing to grok. A current into a cap has a phase shift but not a delay.
Jan
Its a hard thing to grok. A current into a cap has a phase shift but not a delay.
Jan
The current drive simply compensates for the voice-coil inductance's effect. In other words: The driver's intrinsic electo-mechanical lowpass function is reduced by one order. In terms of frequency response (and subsequently phase) its effect is the same as if the the same overall frequency response was achieved by any other means. So from the frequency response view alone current drive doesn't differ much from using EQ. A small difference is that it is automatically correct while an EQ has to be set accordingly. The area where current drive makes the biggest difference is when it comes to the nonlinearity isues of the voice-coil inductance.
I once wanted to try what tmuikku proposed somewhere at the beginning of this thread: Using a series inductor in order to have a low output impedance at low frequencies and a higher one a higher frequnencies.
This way you would still have the damping of fs by the amplifier and some liearising effect (which wouldn't be as great as with pure current drive). But maybe it is still worth the effort. Instead of using a real inductor it could be achieved by the use of frequency-dependant current feedback. The pole-frequency determined by Lvc and Re would be shifted lower and this would have to be compensated for by the use of a high shelf before the amplifier.
Regards
Charles
If the output impedance depends that way on the open loop gain, it means that the higher the open loop gain - the higher the output impence, right?
What if i lower the value of the current sensing resistor by a foctor of 10 and add an amplifier with the respective gain (10) in the feedback path?
What if i lower the value of the current sensing resistor by a foctor of 10 and add an amplifier with the respective gain (10) in the feedback path?
This is not desirable as phase margin will decrease and performance THD, noise floor. But the good side that you will loose little energy/ heat on Rsense resistor. I use 1ohm resistor often. Especially for fullrange efficient driver and real music signal its not that much as it seems that 1ohm. And the conversion is very nice. 1V gives 1A to the load 🙂
Not that hard 😉
I´ve been using mixed feedback in my Guitar amps since 1972 😱 when I tried to mimic better the 100W 4 x 6L6 low feedback amps dominating the scene way back then (think Fender Twin Reverb and similar).
Pencil and paper of course, only Computer I had available then was the IBM 1620 at Universidad Católica Buenos Aires Faculty of Engineering .
I could have used it for design, it could run a crude early variant of SPICE, output was huge "bedsheets" of zig zag folded paper "Line Printer", no CRT monitors involved or existing 😱
Really, it was easier go the pencil and paper route, plus my trusty Staedtler slide rule.
I designed my amps to have output impedance=4 ohm, hence "damping of 1" ,which I had measured in the Tube amps.
What I wanted to know - what does it with the amps output-impedance? The sensing resistor is lower in value and the extra op-amps open loop gain is in series with the amps open loop gain ... so i assume the output impedance will rise a lot, if I am right.
For a motor: F = B*I*L
In the case of a speaker with a voice coil of N turns, it is:
F = B*I*(2*pi*R*N)
Moving in a magnetic field induces a voltage as follows:
V = dΦ/dt
=> V = 2*pi*R*N*v*B
And now, self inductance of the coil! But this, is mounted on a solid iron former, which means, eddy currents, in the iron former cause self inductance effects to be extremely opposed. The effect is much like talking about a transformer primary with the secondary being a shorted thick copper single turn.
Is the residual inductance worthy of consideration?
In the case of a speaker with a voice coil of N turns, it is:
F = B*I*(2*pi*R*N)
Moving in a magnetic field induces a voltage as follows:
V = dΦ/dt
=> V = 2*pi*R*N*v*B
And now, self inductance of the coil! But this, is mounted on a solid iron former, which means, eddy currents, in the iron former cause self inductance effects to be extremely opposed. The effect is much like talking about a transformer primary with the secondary being a shorted thick copper single turn.
Is the residual inductance worthy of consideration?
Basic rule with current-derived feedback is that the amp starts to work like a current source. So it's Zout becomes very high - if loop gain is infinite, it has infinite Zout.
BTW, the comment earlier about dividing the sense R by 10 and adding a gain of 10 in the loop gain doesn't change anything at DC and low frequencies, where you want a current source. At higher freq the loop gain of the amp and opamp will decrease so there the Zout will decrease.
But if you can keep it up within the band where you want current drive, you decrease the dissipation in the sense R with no penalty (except the cost of the opamp).
Jan
Ignore the DC offset, it's in parallel to Rsense, not the load.
I also can't get simulation to align with the numbers he provides. By putting a current source in place of the load, I get output impedance around 420 at 1kHz and 50 at 20kHz. Putting an AC voltage at the input and measuring without load and then with 4 ohm load, I get approximately the same results. .asc attached with GBW and OLG for generic opamp set to LM3875 datasheet typ. values of 8Mhz and 1 Million.
Math is for masochists! That's my quote for the day. I'm glad to hear I'm not the only one who feels that way. 🙂
That said, I rely on estimates all day long. It's always good to have a ballpark number in mind in case the simulator loses its mind and delivers garbage. For precision greater than I can achieve on a napkin the simulator is certainly handy.
Tom
My idea of the extra opamp amplifiing the sense resistors output being part of the loop was wrong - it builds a loop in itself and is not a part of the power-amplifiers loop.
Thanks for sharing the simulation file. Simulation also works fine for me - it helps a lot when building circuits. One can see if a circuit has the chance to work as expected before it is buildt in reality. And it often helps to understand how a circuit works - one can have a look at signals that were otherwise difficult to measure. But as Tom says: always be aware that the simulators output can be garbage....
Bernd
Thanks for sharing the simulation file. Simulation also works fine for me - it helps a lot when building circuits. One can see if a circuit has the chance to work as expected before it is buildt in reality. And it often helps to understand how a circuit works - one can have a look at signals that were otherwise difficult to measure. But as Tom says: always be aware that the simulators output can be garbage....
Bernd
Imagine a black box in series with the speaker return, and an output voltage for the feedback loop.
Now imagine TWO black boxes. One has a 1R sense resistor, the other has a 0.1R sense resistor followed by a x10 opamd circuit.
To a 1st approximation*, you won't see a difference between the two in your circuitry.
* With the caveat on bandwidth etc as noted before.
Jan
An educated reply usually signifies there is some substance in what a replier tries to communicate.
Thanks.