[DDF]

I received my Dac Ah this week and, after correcting the output resistors to 148 ohms (for the +6V supply) and tweaking the offset pot for 3.3V dc at the combined dac output to center the dynamic range, I have to say I was very disappointed. It sounded somewhat distorted and the bass would lose control. Compared to my bog stock MSB Link Dac, the Dac Ah was a real dog.

I took a few measurements to see if something was amiss. Since the bass was out of control, THD at 100Hz was measured first with the unit warmed up (to allow the dc levels to settle at the dac o/p’s):
http://www3.sympatico.ca/dalfarra/Da...h_Warm_THD.gif

then after 20 minutes in the freezer (!):
http://www3.sympatico.ca/dalfarra/Da...h_Cold_THD.gif

Distortion is definitely higher when cold, but nothing that would explain the bad sound, and the THD performance when warm only slightly exceeded the TDA1543 dac spec. For a 6V supplied version, I wouldn’t put heat sinks on the dacs. The files also show there’s a bit of power supply 60Hz, and probably more importantly, 3rd harmonic 180 Hz present, but its pretty far down.

Not shown are the IM measures, averaging about 0.1%. Also of note is that the THD spectrum is rich in 7th harmonic, which can sound edgy.

For reference, here’s the THD of the test file:
http://www3.sympatico.ca/dalfarra/DacAh/100HzDigi.gif

and of the Link Dac (pretty much at the limit of the test file).
http://www3.sympatico.ca/dalfarra/Da...nk_DAC_THD.gif

So, I dug back inside the Dac Ah and turfed the output op amp with the passive mod well described by Kim in another thread. I’m not fond of the dac ah’s o/p stage: it uses a non-inverting op amp with no dc block on the input. Effectively, the +ve and –ve leads are imbalanced by the dac dc o/p, 3.3V. Even the TDA spec sheet shows the better implementation, an inverting op amp with the +ve lead biased up by Vref. My unit was also shipped with AD847 high speed op amps. The original cct called for 5532s, and the 847s are much more particular about layout. For all I know the 847s may be oscillating (no high speed scope at home or I would have checked).

Instead of depopulating the op amp, I connected its output to ground through a 2.4kohm resistor (the 847 likes to see >1kohm), and ran a 3.3 uF good quality poly cap off the dac o/p to the RCA jack. This is going to be used with a Govibe 5, which has an input impedance of 8.75kohm at full volume, and 9.5kohm at min. The 3.3uF was the min size, to avoid premature bass roll-off. Here’s a little excel spreadsheet to show what I mean:
http://www3.sympatico.ca/dalfarra/Da...y_Response.xls

I hooked it back up and instantly: BANG! THIS is what everyone’s been talking about! Before the passive mod, the drums on Kyuss’ “Thong Song” sounded like they were about to explode into tinder, and the strings of the guitar on the Beasts of Bourbon “Not Gonna Try No More” sounded like they were scraping against the fret board. Honestly, pre passive mod, the Dac Ah is a door stop (well, at least mine was). Now: not bad! Mids are forward, but that’s good and the high’s cleaned right up. The thing’s still a bass monster, the DT770 of dacs

but it’ll make a decent dac for work.

Here’s a final THD measure after the passive mod:
http://www3.sympatico.ca/dalfarra/Da...iveMod_THD.gif

The picture doesn’t come close to doing the change justice, but what does stick out is that the nasty 7th has decreased significantly, as has all the “hash” up at the top end. I don’t know if this explains the difference, but it’s a good thing. The pictures also show that about 6 dB output is lost with the passive mod.

In the end, I think I’ll be happy with the dac ah, but out of the box it’s fatally flawed:
-wrong dac output resistors so output clips
-wrong reference voltage bias so it clips asymmetrically (+ve peaks don’t clip at the same levels as –ve peaks)
-very high speed op amps were used in a design proven in with low speed ones, without adding any required decoupling caps on the power pins
-it uses a flat out dumb output stage design

But after fixing all this, it’s a great way to spend $105. The bass is still a bit silly (exaggerated and not that tight), but nothing eq can’t fix.

Dave

 

[Garbz]

EQ can't fix source problems :S

If you are heavily DIY inclined enough you could try building a proper active I/V stage with no feedback. So far I have found nothing touches these designs in my own DAC which has changed I/V stage 3 times now.

 

[DDF]

Quote:

Originally Posted by Garbz EQ can't fix source problems :S

That's a common fallacy. It won't 100% compensate but of course it can help.

Quote:

Originally Posted by Garbz If you are heavily DIY inclined enough you could try building a proper active I/V stage with no feedback. So far I have found nothing touches these designs in my own DAC which has changed I/V stage 3 times now.

The beauty of this is to me is its small size and its low price. I've been designing audio for 20 yrs and for me personally, this isn't worth signicant effort. I'd rather just buy a stellu, mf etc and spend the time tweaking that. But I'm happy with the dac ah for what it is.

Point of my post was to share my findings. This dac has been discussed for a long time and to my knowledge, this is the first evidence that its unstable and the o/p stage is flat out flawed.

 

[Kim Hardee]

Quote:

Originally Posted by DDF Point of my post was to share my findings. This dac has been discussed for a long time and to my knowledge, this is the first evidence that its unstable and the o/p stage is flat out flawed.

Thanks for the great review! Your discussion about the opamp stage was particularly enlightening. It sounds like they're shipping DAC-AHs with random opamps. Some are OK and some are awful. It's probably a good idea to order the DAC-AH with the socketed opamp option for an additional $13 if you're interested in trying different opamps.

It's interesting that you think it's still bass heavy. Others have said that they think it's a little too lean with the passive mod. I personally like the overall balance. Of course this is all dependent on the rest of the system.

Anyway, thanks again.

 

[DDF]

Quote:

Originally Posted by Kim Hardee Thanks for the great review! Your discussion about the opamp stage was particularly enlightening. It sounds like they're shipping DAC-AHs with random opamps. Some are OK and some are awful. It's probably a good idea to order the DAC-AH with the socketed opamp option for an additional $13 if you're interested in trying different opamps. It's interesting that you think it's still bass heavy. Others have said that they think it's a little too lean with the passive mod. I personally like the overall balance. Of course this is all dependent on the rest of the system. Anyway, thanks again.

Hi Kim,
I think its the layout causing some gremlins with faster opamps. I also recently noticed that I could get the dac to pick up radio stations pretty easily, pointing towards other layout issues. I had it plugged into a govibe 5, with the GV5 unpowered and not plugged in, and the dac ah turned off, and the RF pickup was so loud I could comfortably listen to the music from the local AM on my low sensitivity DT770s! Unplugging the power cord to the dac made the rf vanish. Turning on the dac ah made it vanish also.

BTW, the bass is settling down now, its pretty well balanced. I wouldn't call it lean but its enjoyable.

BTW 2: its really heating up when left powered on. I'm going to investigate why, maybe something is oscilating.

 

[Kim Hardee]

Quote:

Originally Posted by DDF /img/forum/go_quote.gif BTW 2: its really heating up when left powered on. I'm going to investigate why, maybe something is oscilating.

That's interesting about the RF pickup. Is that also with the passive bypass?

The DAC-AH does get pretty hot, especially the LM7806 regulator heatsinks. The totally sealed case doesn't help any. After I converted to 8V and put heatsinks on the DACs I drill a few holes in the bottom of the case and the rear panel. I ended up with more and larger holes than shown in my posted photos. I also considered adding some ventilation to the top panel. Then I measured the temperature inside a stock DAC-AH and my 8V modded DAC-AH (both with top panel on) and found that the stock unit was running about 7 degrees C hotter. Therefore, I decided not to add any more ventilation to my modded unit.

Others have claimed that converting to 8V sounds better than 6V, but I didn't do a comparison. I did the 8V conversion at the same time that I did the passive bypass. I do strongly recommend that if you end up doing the 8V conversion that you also add heatsinks to the DACs and some case ventilation. The 8V conversion will also obviously give you more output voltage. I measured 1.35V RMS max with 220 ohm DAC output resistors. With 150 ohm resistors and 6V you are probably getting about 0.9V RMS max ouput. That's not a huge difference as far as perceived volume, but it helps.

Thanks again for your observations.

 

[regal]

Why is everyone going to 150 ohms on the output resistance, I thought we wanted 170 ohm ?

 

[Kim Hardee]

Quote:

Originally Posted by regal /img/forum/go_quote.gif Why is everyone going to 150 ohms on the output resistance, I thought we wanted 170 ohm ?

174 ohms is theoretically correct if you assume the typical 2.3mA full scale DAC current in the data sheet and perfect Vref at 3.3V. However the DAC current may go as high as 2.6mA and Vref will drift with temperature. Therefore, to totally avoid clipping under all conditions something less than 174 ohms is desirable. There is a direct tradeoff, however, between clipping margin and maximum output signal, so you don't want to overdo it. In reality 170 ohms is probably fine since data sheets usually have some margin in the specs.

 

[DDF]

Kim, thanks for the heat warning with 8V. I'm a bit surprised how hot the chassis becomes, even at 6V. In hindsight, I'll be ordering the dac heat sinks.

To open up the chassis, I'll probably source some perforated steel from:
http://www.smallparts.com/products/descriptions/pma.cfm

paint it black, and cut a hole in the lid with this hot glued to the underside.

The Rf pickup was with the passive mod. The 3.3uF poly was soldered directly to between the load resistor on the dac o/p and the o/p rca, so the issue probably occurs even with the active out, but I can't say for certain as I went passive right away. When the power cord acts as an antenna, that screams iffy layout.

regal, I was using the lower resistor to guarantee no clipping over all temp, life etc. Kim was correct, my full scale is about 0.9V which is enough as I'm feeding a headphone amp with it. With 8V you'll be fine at 174 ohms.