Transistor SOA gets exceeded when speaker phase is high while impedance is fairly low, then high currents can flow at the zero crossings, while Vce is high. Typically this condition occurs at the upper end of the reflex port tuning. ZZ Top are good at finding this frequency.
A 45V rail amplifier producing 200W into 4R, maybe 3R minimum at port tuning, means maybe 10A@30 degrees, so 5A at zero crossing, so average is 2.5A at 45V Vce. You are looking this SOA at dc ideally.
A 45V rail amplifier producing 200W into 4R, maybe 3R minimum at port tuning, means maybe 10A@30 degrees, so 5A at zero crossing, so average is 2.5A at 45V Vce. You are looking this SOA at dc ideally.
silly question. Does anyone actually know what the impedance of an ESL without the step up transformer is? Obviously capacitive but by how much, and does it change with drive level as the film moves?
It's about a nF for a square meter at the usual spacing. My design goal is 1.5nF at 1500Vrms..
Edit: the way I interpret this is that you have two caps, since the stator between the moving diaphragms is grounded for AC. So this equation gives the total capacitance of two caps, one on each side of the stator. This is the cap reflected trough the usual xformer to the driving amp.
With direct drive, each cap is driven out of phase from a separate output stage so each output stage sees half the cap value from the equation. Have I got that right?
Jan
Attachments
Last edited:
Hi Guys
In output stage design, using the DC SOA is not "brute force" rather pretty much common sense. "Finesse" would be in how you reduce THD and maintain Vq over temperature variations.
With the dual-stator ESLs you have the correct notion regarding the capacitance seen by each amplifier.
Billshurv: The impedance follows the capacitive reactance formula for frequency, so rises at low-f and decreases at high-f. A solid-state output stage should be able to drive 90-degrees, i.e. full current at zero volts out without dying. The protection circuit is essentially just a hard current limit.
Note that the music power percentages that have been studied, and posted by compnaies like Crest, are useful for the design of multi-rail amps inasmuch as setting the rail voltages is critical to heat sink design. Just after the rail switches the instantaneous power dissipation peaks. Few if any of the class-G and class-H amps have very good THD so it is unlikely you will use them in your living room. For PA use, they are economical designs and are combined with spectral compression to maximimse SPL and limit dissipation.
Have fun
In output stage design, using the DC SOA is not "brute force" rather pretty much common sense. "Finesse" would be in how you reduce THD and maintain Vq over temperature variations.
With the dual-stator ESLs you have the correct notion regarding the capacitance seen by each amplifier.
Billshurv: The impedance follows the capacitive reactance formula for frequency, so rises at low-f and decreases at high-f. A solid-state output stage should be able to drive 90-degrees, i.e. full current at zero volts out without dying. The protection circuit is essentially just a hard current limit.
Note that the music power percentages that have been studied, and posted by compnaies like Crest, are useful for the design of multi-rail amps inasmuch as setting the rail voltages is critical to heat sink design. Just after the rail switches the instantaneous power dissipation peaks. Few if any of the class-G and class-H amps have very good THD so it is unlikely you will use them in your living room. For PA use, they are economical designs and are combined with spectral compression to maximimse SPL and limit dissipation.
Have fun
full current at 90 degrees would be a very wasteful design.
Nearly all amplifiers have some degree of foldback current limiting to keep inside the SOA.
Practical loudspeaker rarely go beyond 30 degrees while the impedance is low.
Quad ESLs were a special case
Quad ESL-63 loudspeaker Measurements | Stereophile.com
The Stereophile measurement did not have a phase curve, but the resonance at 22kHz is so high Q it must be crazy
It could be argued what Jan is doing is NOT a practical loudspeaker. The stereophile curves are for a QUAD with its transformers in place, so a somewhat different load to going straight in. Hence the questions. Like many the idea of direct drive ESLs has tempted me over the years but always stopped before I looked too deeply as danger to life is significant and I'm not experienced at things over 1000V.
Am I right in assuming RMS currents are around 10mA?
Hi Guys
davidsrsb: The Quad has an input transformer, like all commercially available ESLs. Jan was talking about direct drive of the stators.
If you look at the 1970s Audio Amateur mags, there are a couple of ESL drivers using vacuum tubes. This is the natural choice since tubes are king when it comes to high voltage. However, these days there are mosfets rated for 4kV and cascoding is an option for using 1kV devices.
LETHAL VOLTAGES are present in any full-scale ESL or driver circuit.
It would be best to locate each amplifier within the speaker housing so that these very high voltages are contained and not run through cables from another box. This is convenient inasmuch as balanced signals can be fed to each speaker using XLR connectors and each of the four amplifiers is identical and relatively independent.
Yes, currents are fairly small but not zero.
Have fun
davidsrsb: The Quad has an input transformer, like all commercially available ESLs. Jan was talking about direct drive of the stators.
If you look at the 1970s Audio Amateur mags, there are a couple of ESL drivers using vacuum tubes. This is the natural choice since tubes are king when it comes to high voltage. However, these days there are mosfets rated for 4kV and cascoding is an option for using 1kV devices.
LETHAL VOLTAGES are present in any full-scale ESL or driver circuit.
It would be best to locate each amplifier within the speaker housing so that these very high voltages are contained and not run through cables from another box. This is convenient inasmuch as balanced signals can be fed to each speaker using XLR connectors and each of the four amplifiers is identical and relatively independent.
Yes, currents are fairly small but not zero.
Have fun
It is low frequency content that will get you and an input blocking cap or
filter makes a difference.
I stumbled across these old post that you might find interesting:
http://www.diyaudio.com/forums/soli...terview-error-correction-231.html#post1310838
http://www.diyaudio.com/forums/soli...terview-error-correction-231.html#post1310862
I have some thoughts that I'll post when I have more time but some food for thought for now.
filter makes a difference.
I stumbled across these old post that you might find interesting:
http://www.diyaudio.com/forums/soli...terview-error-correction-231.html#post1310838
http://www.diyaudio.com/forums/soli...terview-error-correction-231.html#post1310862
I have some thoughts that I'll post when I have more time but some food for thought for now.
- Status
- Not open for further replies.