Anyone interested in a digital amplifier project?

[MWP]

Originally posted by koldby

Wonder if anybody here has seen this:
http://listen.to/audioexperiment
I know this is not a digital in amp, but it sure is a DIY project!!

Errr.... what is it?

[koldby]

Hi MWP

It is a self oscillating class D amp. , but an analog in design....


Koldby

[dwk123]

There may not be any point in reviving this thread, but I recently became very interested in the Panny XR45, initially due to this editorial from the Newform site:

http://www.newformresearch.com/updateaug03.htm

After doing a bit of research (including this thread), I'm very interested some ideas similar to what Brian is doing. Specifically, I'm interested in trying to tap into the I2S signals to/from the equibit processors and bridging them into an I2S based soundcard like the Envy24 cards. This would allow both direct digital output of the multi-channel signal to the PC, as well as direct digital output from the PC into the amp sections.

I (of course) know just enough about all this to be dangerous, but any updates on experiences with the Panasonic units or other equibit (or other) amps along these lines would be greatly appreciated.

[Brian Brown]

I’ve done some more investigation and experimenting with my SA-XR10. Here’s some new information that I can share:

The main Equibit amplifier board appears to be made by Texas Instruments. It seems that Panasonic just built the rest of the receiver around it. This board is pretty well self-contained. I’m able to operate it independently outside of the receiver. The layout is very similar to TI’s other Equibit reference boards (and it’s much cheaper to buy the receiver).

Panasonic sells the service manual for this receiver. It includes full schematics.
Order number: AD0207128C1
1-800-833-9626
They only charge about $7 for it. It’s fully worth the price for the detail that it provides, rather than try to make copies.

I can offer the following supplemental information:

Most of the connections to it are made with ZIF connectors. The bare end of stripped wire can be inserted into them and locked into place.

These include:
Five Speaker Outputs (channels 1-5) (the receiver adds some snubber caps at the final output terminals).
Subwoofer line out (an op-amp circuit integrates the channel 6 Equibit PWM back into a line signal).
Interface to headphone output circuit board (can be left unconnected if using the board outside the receiver).
Main bus power (45VDC nominal).

The main input to this board is through a 25pos 1mm FPC connector.
It includes the digital audio interface, status/control lines, amplifier power supply control, and low level power.

PINOUT <direction> NAME (description):
1. <digital_in> MCLK (256fs)
2. <digital_in> SCLK (64fs)
3. <digital_in> LRCLK (fs)
4. DGND
5. <digital_in> FRONT DATA (I2S 24bit data)
6. <digital_in> SURROUND DATA (I2S 24bit data)
7. <digital_in> CENTER / SUB DATA (I2S 24bit data)
8. DGND
9. <analog_out> DEC (Equibit bus voltage feedback) (DEC = BUS / 11)
10. DACGND (DGND)
11. <digital_in> DOUBLE SPEED (for audio data rate) (0: fs = 44.1KHz or 48KHz, 1: fs = 88.2KHz or 96KHz)
12. <digital_out> /TEMP WARNING (0: High heat sink temp – amp still functions, 1: Normal)
13. <analog_in> VOLUME (Equibit bus voltage control) (BUS = VOLUME * 13.3)
14. <digital_out> /Shutdown (0: Overtemp or output short – amp disabled, 1: Normal)
15. <analog_in> AMP RESET (see note)
16. AC GND (DGND)
17. <digital_in> /MUTE – HeadPhones (0: Headphones disabled, 1: Enabled)
18. <digital_in> /MUTE – Front (0: Front speakers disabled, 1: Enabled)
19. <digital_in> /MUTE – Center & Surround (0: Center and surround speakers disabled, 1: Enabled)
20. <digital_in> /MUTE – Subwoofer (0: Subwoofer line out disabled, 1: Enabled)
21. 12V_GND (DGND)
22. +B (12V)
23. 12V_GND (DGND)
24. +B (12V)
25. <digital_out> Headphone Switch (0: Headphones plugged in, 1: No headphones)

NOTES:
Logic signals are 3.3V (LVTTL).
This board derives its internal 3.3V supply from the external +B (12V) supply.
The AMP RESET input is not a logic control signal. It is a +5V supply connection to a power monitor chip. It is intended to be connected to an external +5V supply that is derived from the same +B (12V) supply that feeds this board.

************

This board is based on three stereo TAS5012 Equibit modulator chips. The power output stage is a discreet implementation. It has five channels of amplification. The sixth (subwoofer) channel goes through the same Equibit modulator section as the other five channels, but instead of proceeding through a switching power output stage, it gets integrated into a line level output.

The board also has a switching power supply to provide the variable-voltage DC bus for the Equibit output power H-bridges. This is based on a UC3849DW Secondary Side Average Current Mode Controller. This power supply sets the bus voltage at 13.3 times the voltage input to the board’s VOLUME input. The bus voltage can be verified by reading the voltage of the DEC output (bus voltage divided by 11). Lowering the bus voltage decreases the output volume without reducing the 24bit audio resolution.

Variable bus voltage can achieve 20dB of output level variation. The variation is limited to 20dB because of two constraints: The minimum voltage is limited by the voltage drop of the output MOSFETs and the output low-pass filter. The maximum voltage is limited by the voltage rating of the output MOSFETs and/or the gate driver circuit. The SA-XR10 has a bus voltage range of 4.33V to 42.2V (20dB).

The output volume is controlled by a combination of digital attenuation and variable bus supply voltage. This board requires external control of the bus voltage and some type of digital attenuation before data is input into it.

The scheme the SA-XR10 receiver uses is as follows:

0dB (full output): no digital attenuation and full bus voltage.
Note that (assuming the data isn’t clipped digitally) this amplifier can’t be driven into clipping. (It is possible to cause the thermal shutdown circuit to activate, however.)

The first 20dB of volume control (-1dB to -20dB) are done digitally. (At -20dB, the digital resolution is effectively reduced from 24bits to a little over 20bits).

The second 20dB of volume control (-21dB to -40dB) are done by reducing the bus voltage. (The effective resolution remains at about 20bits).

The remaining 39dB of volume control (-41dB to -79dB) are again done digitally. (At -79dB, the effective resolution decreases to about 14 bits – pretty good considering that the peak levels at this setting are really quiet.)

Below -79dB the output is muted.

To operate this board without the rest of the receiver, you have to supply your own combination of digital attenuation and/or level control.

A very crude initial method that I’ve been successful with is to shift the input data right 3 bits (-18dB) with a CPLD and then use a pot to control the VOLUME supply control input. I use a 10K log taper pot with the top side connected to 3.3V through a 453ohm resistor and the low side connected to ground through a 1.13K resistor. This produces a log taper control voltage at the wiper with a range of 0.3V to 3.2V, which corresponds to a supply voltage range of about 4V to 42V. This simple control method lets me have a range of -18dB to -38dB, which has been completely adequate.

*********


I view the variable switching power supply as the weak point of this board’s design. Watching the bus with a scope, things are pretty clean at lower levels. At high output levels things start to get pretty wiggly. The Equibit output section basically has no power supply rejection, so this is an issue. (I’m sure this is the reason for the SA-XR10’s published 0.9% THD at 100W, it will be drastically lower at reduced power levels.) The Equibit design really needs a stiff, low-impedance supply even more than most other amps. As an initial experiment I isolated the bus from the switching supply and hooked it up to three 12V car batteries in series. This made an astounding improvement at loud levels. I’ve already mentioned the surprising clarity and imaging this unit had at more moderate levels. With the batteries, percussion was especially improved.

**********

I’m still concentrating on a multichannel digital interface for DVD-A and SACD at the moment. I’ve been doing some experiments with different clocking and reclocking schemes. I’m not ready to report on this yet, but I’ve made some good progress. I’ll (hopefully) be getting back to my Equibit experiments soon.

(Did anybody else notice that Digikey now has TAS5182 chips available?! TI hasn’t even been sampling them yet.)

*********

A couple of comments about the SA-XR25 and SA-XR45:

I haven’t seen or heard either one of these in person, much less probed around inside of one.
I have reviewed some of their documentation.

It appears that both have identical Equibit sections.
There are some major differences from the SA-XR10:

The Equibit board also has a local processor that supports other receiver functions.
The bus supply is on a separate board and operates at a single voltage.
Digital attenuation is done in combination with a TAS5036 six channel Equibit modulator chip (with built in volume control).
There are six power output stages instead of five.
The output stages are now based on the TAS5182 chip, instead of discreet components.
The output MOSFETs and low pass filters appear to be identical to the SA-XR10.

I’m not surmising that these are any better or worse in audio quality than the SA-XR10. I just wanted to point out that interfacing to the internal boards will be quite different from what I’ve reported above.

**********

I’ll post some more stuff later as I progress with my (slow) progress.

Regards,
Brian.

[fcserei]

I'm using the SA-XR10 now in a fully digital i2s connected chain (transport -> ultracurve -> sa-xr10) with full range horn loaded Stax electrostatics (50-20k) on the LR channels and an 18" ELF woofer on the sub channel (10Hz-50Hz). In this setup the built in cyrus decoder vorks fine as a 2 way crossover.
I'm also thinking about power supply modifications or battery supply but I have a question.
Could you figure out how the volume control works on the sub and headphone channel. They are both integrated back from the output of the equibit chips and according to your report the digital volume control made on the DSP board does not shift the bits in the -20 - -40dB range. But it seems both channel follows the volume control even in this range.

Do you have an idea what is the max supply voltage the unit can handle?

Thanks

[JOE DIRT®]

Great post Brian!!....I`m going to do some reading on this...you have me fascinated

CHEERS!!The DIRT®

[dwk123]

Seconded - thanks for the info.

I opened my XR25 up, and it does look rather different from what you describe. The input/decoding board looks pretty densely packed with stuff, but it *looks* like the interface down to the Equibit board might be pretty straightforward.

One big-ish difference is probably going to be in the sub channel handling. Since there are 6 powered output channels in the XR25/45, the sub does not appear to go through an Equibit stage, although from a casual observation I can't see how it is handled.

I have two goals - short term and long term. Long term, I want to get to a full 6-channel I2S interface. In the short term, though, I'd be happy just getting use of all 6 channels from an external analog in.

Brian, any suggestions on where to get the order number for the XR25/45 service manuals. I suppose I can just call the number and wing it.

[jwb]

I'm interested in experimenting with this topology for a variable-output SMPS. It's from Linear AN70.

[Brian Brown]

Originally posted by fcserei
Could you figure out how the volume control works on the sub and headphone channel. They are both integrated back from the output of the equibit chips and according to your report the digital volume control made on the DSP board does not shift the bits in the -20 - -40dB range. But it seems both channel follows the volume control even in this range.

For the subwoofer, the volume control is done strictly by digital attenuation, even in the -21dB to -40dB range.

The headphones are integrated from the same Equibit PWM stream as the front speakers. The SA-XR10 can't drive Speakers and Headphones at the same time.

Please note that the Headphone Switch (Pin 25) description I gave in my previous post was backwards. It should have read 0: No headphones, 1: Headphones plugged in.

When *no* headphones are plugged in, the Headphone Switch signal is low. The receiver then pulls the /MUTE - Headphones line low (disabling the headphone integrators). The /MUTE – Front line is high (enabling the power MOSFET stage for the front speakers). Volume control is done as I previously described.

When headphones *are* plugged in, the Headphone Switch signal is high. The receiver then pulls the /MUTE - Front line low (disabling the power MOSFET stage for the front speakers). The /MUTE - Headphones line is high (enabling the headphone integrators). In this state volume control is done by digital attenuation only.

Do you have an idea what is the max supply voltage the unit can handle?

The IRFIZ24N output MOSFETs are rated at 100V. The flyback voltage that these devices see could peak out at double the bus voltage. Also, the bus filter caps are rated at 50V. These two factors would indicate a maximum of 50V. Personally, I wouldn't try to go higher than the 42.2V volts originally designed into the unit.

The real limiting factor of this output stage appears to be the thermal dissipation of the MOSFETs. I don't think a higher bus voltage would be of value unless a completely new output stage was designed.

Brian.

[Brian Brown]

Originally posted by dwk123
One big-ish difference is probably going to be in the sub channel handling. Since there are 6 powered output channels in the XR25/45, the sub does not appear to go through an Equibit stage, although from a casual observation I can't see how it is handled.

On this unit the sub line outputs and the headphone outputs are handled by conventional A/Ds. They are completely separate from the Equibit section.

Brian, any suggestions on where to get the order number for the XR25/45 service manuals. I suppose I can just call the number and wing it.

The manual for the SA-XR25 is order number MD0302055C1.
I don't know the number for the SA-XR45, but you should be able to get it from Panasonic Customer Service (1-800-833-9626) by asking for it by description.

Brian.