Go re-read Hugo's post!

He specifically stated that the Rio (Mk.2A) player has these problems fixed. CSR, as you pointed out, reviewed the empeg (Mk.2) version. We were so shocked to see the figures in that review that we went back to the manufacturer and discovered some production problems (that were not present in the original design). Hugo described these in his post.

Rob

CSR's technical guy in Florida tooks his measurements using (as I recall) reasonably high bitrate MP3 files. There wasn't any straight forward way to play WAV files at that time, and in any case they made the point to me that they were testing a whole product - which was specifically intended for MP3 playback. That's reasonable.

Rob

OK, Hugo...

If you're going to be popping off about the MK2's need part replaced for home outs then I need to know which it is (sn 080000404), also which caps need removal to change that roll-off?

Thanks in advance for the info... :-)
Tim

I have measured the frequency response of my MKIIa and see some rolloff at both ends of the spectrum. It's not too bad when playing a WAV off the EMPEG, but is a bit more significant when playing a test source through the EMPEG (via the aux-ins).

The attachments to this message (and the next two replies) show the frequency response I measured on my primary MKIIa a few weeks ago when I was setting up my PC based RTA. These graphs already have a compensation curve factored in that compensates for my sound card's response anomalies.

The RTA resolution is set at 1/12 oct in all these images.... Curve smoothing is not enabled, which is why you see steps in the response curve on 1/12 octave boundries.

The image attached to this message shows the frequency response measured at the EMPEG's 1V outputs while playing a pink noise WAV ripped directly from directly the 2001 IASCA test disc.

The response is pretty good, but other audio equipment I have tested is generally flatter.

I tried various output levels (-10 dB to 0 dB), the 1V outputs, the 4V outputs... all yielded pretty much the same results.

Looping my sound card output back into it's line input results in a nearly ruler flat measured response with the same compensation curve (see second reply to this message). Playing this pink noise track on a good home audio CD player also yields a flat response curve. Therefore, I am confident in my test setup, and that these graphs shows the actual response of my EMPEG.

And this attachment shows the frequency response through the aux inputs. Basically, the PC's audio output is fed to the EMPEG's aux inputs. And then the EMPEG's outputs are fed back into the soundcard.

For reference I replaced the EMPEG in this test chain with a set RCA barrel connectors, without changing anything else in the test setup (so that the input and output patch cables were still in the loop). With the EMPEG barreled out of the chain, the response was flat from 20Hz to 20kHz. The measured response of this test is in the next reply.

And lastly, this attachment is a test of my measurement system. This is with the PC sound card's output externally looped to the line input of the same card. Basically, I just disconnected the RCA patchcords from the EMPEG, and put RCA barrels in it's place (so that the input and output patchcords are still in the test path).

This also shows that the slight level offset between the left and right channels in the previous graphs is partially due to my sound card.

Yes, and it was this review which caused us to go "hang on a minute! the original production units were much better than that"! We did contact them about this, ISTR we were going to send them a later unit and they were going to re-do the audio tests. Actually, we were a little annoyed that they didn't ask us if we were aware of this issue before publishing the figures, because 40dB is pathetic and does indicate a design/manufacture problem.

ISTR we were down something like -0.5dB to -1dB at 20kHz on later mk2s and all mk2a's.

The stereo separation problem was due to a 100 ohm being stuffed in the car output circuit instead of 0 ohm - again, this was fixed on the production line & any units that came back to us had it done as a FCO. This, as you say, is a biggie - it's NOTHING to do with the aux input specs (which *does* have left/right interactions as the high-CMRR front end takes a ground input common to L/R to reference both from - it's likely that there will be some crosstalk introduced here).

Hugo

I seem to recall that we did get the results before they printed them. I supplied a great deal of feedback to their copy and they integrated some of that into the review. They chose to print the test results even though I refuted them.

The moral of the story is to always check review units thoroughly before sending them out, because you don't get a second chance (a lesson taught to Rio by the WSJ recently).

Rob

Any updates on the progress of the project? I have an oppurtunity to purchase a very good D/A converter for Car Audio use but, it is not cheap...even though it is used. Just looking for a little head's up on this.
Dave

Talking about volume control:
Controlling the volume on the digital path is quite a non-sense operation.
This will reduce the digital resolution down from the original 16 bits. This means reduce the Dynamic and increase the background noise.
The digital output of the SoundBlaster works in this way... when i use it, i have to set the PC volume to the maximum before clipping and set the volume of the ampli to get the proper volume level.

The only way to avoid this problem on the EMPEG , is to use a ANALOG volume controller connected after the external DAC... it could be a active or a passive preamp.

I hope, anyway, that the digital output will be a fix-volume one… just after the CPU, like any CD player has.

P.S.
I do not think the EMPEG sets the volume on the digital path ... There is probably some where ... in the analog path of the signal, a gain controller. It could be inside the dac.

Ok, the update. Well, I have all the parts I need to build my first prototype with the exception of the PCB, which I ordered last night. This should be in by Friday. So, I should be able to build and test the circuit this weekend. This protoype is not surface mount, that will be a few more weeks. If you can't wait, I may be able to get a couple out in the next two weeks, but these will take up the second drive bay. The surface mount version will fix this problem. Before I go to work developing that however, I have to work out any bugs in the original design.

Either way, these will have a coax S/PDIF output and that's it. You can't please everyone.

Stu

Great work. Definately getting one if you can get all the bugs sorted. Any ideas on pricing yet?

Now what would be neat would be digital in...

- Trevor

No, it does volume control digitally - lower volumes *will* loose resolution. It does dithering, etc, internally. This, as you say, is very common in highly integrated digital audio solutions. There is a curve on the SAA7705H specsheet which shows the THD+N curve vs volume setting (page 15 on mine).

Note also that the DACs are 18-bit (and the internal path is 18 bits wide), which gives us 2 bits of attenuation (6dB) without any loss of resolution. You'd be looking at a very low noise external attenuator (eg, LM1972 as used in the Rio Receiver) with good PCB layout to give an analogue volume control with less ill effects than the existing digital one.

Some DACs, eg the PCM1716 in the Rio Central/Rio Receiver, are 24 bit, allowing 8 bits of attenuation without loss of resolution on a 16-bit signal.

Hugo

Regarding pricing, I had said earlier that it would probably be $100 US. This is still a reasonable figure depending on actual interest. Kits would run $65 or so. The more PCBs I make at once, the cheaper it's going to be. Same can be said for the other parts, but not to the same extent.

Stu

Once I get the board built, it's going to need 5 volts to power it. You say that's available on the Empeg board and indeed I did see a couple headers labeled 5v with jumpers installed in them. Do these jumpers carry 5 volts from one section of the board to another? Could I get the 5 volts by installing a Molex type connector on one of these headers? What size are they? Thanks.

Stu

The first board has been assembled. I'm going to be brave and hook this sucker up to my RioCar later today, after I get some sleep. Still to be done is to find a good source of +5 volts and the proper settings on the chip. Luckily this version of the board has jumpers. I'm not quite sure how to handle the reset pin on the chip. I left it floating. Hopefully it will work that way. I couldn't tell from the documentation whether it was necessary or not. I'll post the results when they are in.

Wish me luck.

Stu

Everyone of us wishes you all the luck in the world!

Dalsemi/Maxim make some uP reset chips which might do the job if you do need a proper reset on power up. I've used the EconoReset series and they work quite nicely.

- Trevor

Yep, the empeg itself uses a MAX809T

Hugo

Update on the SMT board:
The design of the PCB is done and is now roughly 2.75" (6.99 cm) x 1.09" (2.78 cm) and that's where its going to stay. This is down from the original board's dimensions of 3.81" x 2.50" or a 68% reduction in footprint, even with additional circuitry. The parts are on order. I'll check for fit and order up the minimum quantity of PCBs.

Stu

I have decided to make the board capable of outputting coax, Toslink, and I2S. The basic model will come equipped with coax only at a price of around $100. Other models will include an additional connector and cable for Toslink, I2S, or both. So the maxed out version will offer all three formats on one board at a price of somewhere around $135. The Toslink/coax board will be priced around $130 and is only going to work on Rios/Empegs with one drive bay though, as the Toslink connector is too deep to allow for the 2nd drive, but the option is there. I2S is untested, so I make no guarantees as to whether this will work or not. The I2S/Coax board will be around $110. The coax will be via RCA phono jack. It is probably possible to route the signal out of the mic in cable for docking purposes if need be, but I haven't explored this option yet. I'll let the end user decide how to handle that.

Parts should be here the first part of this week, so I should have the PCB by Friday.

Stu

Sounds good.

For those interested. We finally got the digital connectors and board permanently installed. Here is a pic of the rear of the player.

Stu

Here is a pic of the other side of the rear panel with everything stuffed in there. It's a very tight fit, but at least it fits.

Stu

Nice work!

Hmmm, just how necessary are the connectors? Are they needed to get everything in place or could one solder the wires directly to the board or a row of pins or something like that? That'd, as far as I can see, give the disk plate a bit more room to move...

/Michael

Thanks. Believe it or not, there really isn't any interference with the drive or sled movement, but yes, the wires could be soldered directly to the board. In fact, that's what we had to do for the optical wires to fit.

Stu

Hmm, really impressive.

TommyE

vfd's bet outside of tables. microphones, states, and wolves 10 miles from mice poopoo by insects. feet or tables give up.

For anyone curious, the first edition of the install manual for the board is online. Creating It was quite an arduous task. We hope you find it useful. Get it at www.eutronix.com/media/dig-cxxxx_manual.pdf . The photos within are very large and as a result its a hefty download at around 5.5 MB.

Stu

Update:
The shipment of optical modules has arrived. All that remains is the PCB, which we expect to have in another 10 days.

To those of you who have downloaded the manual, what kind of download speeds are you getting? I'm just curious, because there really isn't any way to change it anyway.

Thanks,

Stu