John Curl's Blowtorch preamplifier part III

[31697B]

I am associated with the design of a power amp that uses switching amps (Bruno's) as variable power supply rails. It works pretty well, BUT it has too much residual RFI generation for my taste. It isn't even measurable without extra filters. I can carry it under my arm, and it has about 300W/ch. Still I prefer pure linear Class A circuitry. (Fets preferred) I am working on a 'better' power amp, but it will be too big for my living room, but it will be almost all fet (except for the output stage) and more Class A than the JC-5. More expensive too!

nice.


-RM

 

[wayne]

Did an amplifier like this at Threshold about 1990 for one CES. Digital in with a 20 ms ram delay while the switching supplies ramped up. Not so good if you charge by the pound.

 

[Mark Johnson]

I bet the "RAM delay" was actually a FIFO.

1 kiloword x 16 bits x 2 banks (left, right) would do the job since (1024 / 44100) = 23 milliseconds. 1K x 2bytes x 2 = 4 Kilobytes. Carrumba.

 

[31697B]

We tried SMPS recommended over here and the noise was excessive IMO. The high peak currents needed seemed an issue at this level of power. At some point the costs of additional filtering and complexity and potential life and reliability/repair costs etal… Just didn't make it. The Benchmark does it at a cost for modest (to me) power. Someday. The weight and cost to ship heavy gear is a significant added cost.



-Richard

 

[wayne]

I bet the "RAM delay" was actually a FIFO.

1 kiloword x 16 bits x 2 banks (left, right) would do the job since (1024 / 44100) = 23 milliseconds. 1K x 2bytes x 2 = 4 Kilobytes. Carrumba.

Yes it was a FIFO, actually fairly expensive back then if I remember correctly.
A 8 bit dac fed the swps reference with a hold cap it was kind of crude but worked.

 

[Wavebourn]

Hi from California Audio Show!

 

[simon7000]

Hi from California Audio Show!

Scary!!!

 

[Wavebourn]

Scary!!!

Oops... Is the forum hosted in Australia?

 

[miklos]

No. That's what happens when you take a picture with a telephone and not paying attention to the up.

 

[canyoncruz]

Did an amplifier like this at Threshold about 1990 for one CES. Digital in with a 20 ms ram delay while the switching supplies ramped up. Not so good if you charge by the pound.

I think I used that amp at the 94 CES

 

[simon7000]

Thumbnail is upside down but displays correctly when selected for me.

 

[canyoncruz]

Could it be SMPS design isn't up to snuff in audio? They do amazballs stuff in industrial, even analog stuff

 

[DPH]

Technically or in the court of public opinion?

 

[canyoncruz]

Technically

 

[DPH]

Yeah, it's not like the rest of the world has not moved on, including infinitely more sensitive equipment than audio gear.

So technically limited seems a bit rich. Poor implementations sure, then again so are poor linear supplies.

 

[Max Headroom]

Oops... Is the forum hosted in Australia?

Your pic displays upside down here too.....the problem is your side of the equator .

 

[Wavebourn]

No. That's what happens when you take a picture with a telephone and not paying attention to the up.

So, it is not about Equator... Thanks!

 

[31697B]

The smps is best used for steady current draw loads. There they work fine; computers, many test equipment etal. Steady load current.


In preamps, a simple small transformer etc and three term reg is lower cost than smps. and quiet. And put no switching noise onto the ac line.


Its just a bunch of trade-offs to determine which to use.




THx-RNMarsh




.

 

[Markw4]

The Benchmark AHB2 power amp uses an HF switching supply. It has incredibly good specs and is on the Stereophile recommended equipment list.

 

[chris719]

The smps is best used for steady current draw loads. There they work fine; computers, many test equipment etal. Steady load current.


In preamps, a simple small transformer etc and three term reg is lower cost than smps. and quiet. And put no switching noise onto the ac line.


Its just a bunch of trade-offs to determine which to use.




THx-RNMarsh




.

Computers are hardly steady current draw. Everything in the last 10 years is aggressively designed with power gating and a race-to-idle strategy.

My Core i9 7900X CPU alone uses around 12W at idle and peaks at something on the order of 270W (overclocked). This step occurs quickly, on the order of us to ms depending on CPU load state. Out of the 1.8V multi-phase buck converter that feeds the CPU that is a step in current from ~7A to 150A. There are various settings on these motherboards for controlling transient droop / overshoot. The 1.8V regulator is using 12 ISL99227-27B (60A rated each!). And that's not even counting the graphics card which uses something like 16W idle and tops out at 220W in 3D rendering or a game.

Have you heard laptops and PCs sing? If you are able to disable all power management and run the CPU at full clock and voltage you would notice the coil and capacitor whine basically goes away.