Here is an interesting graphic that you may not have seen. The AKM dacs feature a 2v voltage output and super low ultrasonic noise floor which enables them to run their analog output direct out to the output connector with no analog filters or buffers, still providing plenty of power to drive passive stepped attenuators. No active circuitry of any kind following them. This provides the purest, most detailed sound that I have heard. The direct out mod is very cheap to put into a DEQ2496 or various sound cards that use the AKM dacs needing only a high quality coupling cap to block the dc offset. Interestingly, I find that the sound of the 4395 has better fine resolution but the 4396 has a beautifully focused and powerful presentation.
Attachments
Thanks Scott, very nice!
You seem to be not alone in your preference for the '95 over the '96.
I have found that good chips like these need only a transformer and a resistor on the outputs for spectacular results. 😀
You seem to be not alone in your preference for the '95 over the '96.
I have found that good chips like these need only a transformer and a resistor on the outputs for spectacular results. 😀
Hello, nice graph...
Do you have any examples of such transformers?
I have found that good chips like these need only a transformer and a resistor on the outputs for spectacular results.
Do you have any examples of such transformers?
Direct through caps
I tried some nice Sowter transformers but found that I like the sound direct out from the dacs through Dayton foil caps the best. The trannies did have a nice liquid midrange though.panomaniac said:I have found that good chips like these need only a transformer and a resistor on the outputs for spectacular results. 😀
yes I have heard of sowter being used in this position with akm parts. I was wondering about some less expensive transformers..
jwb said:What kind of test signal is used?
To me it looks like 1kHz sine at -60dBFS and the measurement done with few FFT points or possibly a window choosed for truer represenatation of noise rather than static signals.
/Peter
Sowters are the cheap trannies
The Sowters ARE one of the value leaders at $120.00. Other high end transformers go way up from there. $200, $300 each! That's why I like the cap coupling method. With the right caps, Dayton foil, it sounds better and they only cost $15.00SunRa said:yes I have heard of sowter being used in this position with akm parts. I was wondering about some less expensive transformers..
I just like the sound of transformer coupled circuits better than caps. A matter of taste, I suppose. But many who have heard my transformer DACs like them, too.
Sowter has been mentioned, and of course there is Lundhal. Shootz! I've even used Edcor - hard to beat for $5.
My favs are still the Onectics, custom made by our own Mr. Bud Purvine, "Transformer Gypsy" and EnABL guy.
As opposed to a lesser transformer, they are much more open and dynamic.
I should have a pair of Jensen transformers here this week for testing. I'll let you know.
Remember, it is a matter of taste. Something as simple as DC blocking caps may float your boat, as sendler mentions. Whatever works. Just keep away from those opamps.
Sowter has been mentioned, and of course there is Lundhal. Shootz! I've even used Edcor - hard to beat for $5.
My favs are still the Onectics, custom made by our own Mr. Bud Purvine, "Transformer Gypsy" and EnABL guy.
As opposed to a lesser transformer, they are much more open and dynamic.
I should have a pair of Jensen transformers here this week for testing. I'll let you know.
Remember, it is a matter of taste. Something as simple as DC blocking caps may float your boat, as sendler mentions. Whatever works. Just keep away from those opamps.

I'm using my AK4396's analog output pins directly without any filter/opamp/buffer etc. Just a 1K resistor between my Gainclone amp. I'm not very experienced listener but I don't know how could it sound better... Maybe tweaking with power supply section and changing oscillator for low jitter. That's all.
Resistor
I assume you have a blocking cap on the input of the amp? Why the 1k series resistor?I'm using my AK4396's analog output pins directly without any filter/opamp/buffer etc. Just a 1K resistor between my Gainclone amp. I'm not very experienced listener but I don't know how could it sound better... Maybe tweaking with power supply section and changing oscillator for low jitter. That's all.
why does everone seem to think oder 1 million cmos digital transistors and a internal cmos analog output op amp built on the compromised mixed signal process is somehow "perfect"
and immediately outside that hugely integrated mixed signal chip it is suddenly time to "get simple" to not "spoil" the sound
particularly comparing the adc internal cmos output op amp to today's fast, extended input linearity chips like AD4899 with Gilberts tanh compensated diff pair input built on a dedicated linear semi process
or any of the past decade's precision fet input op amps with 20-60 MHz GBW - some aimed at the opa627637 - or just put out the $ for the opa627
these chips won't have trouble in a properly designed I/V application - should perform far beyond NE5534 and the adc's internal analog circuitry
and immediately outside that hugely integrated mixed signal chip it is suddenly time to "get simple" to not "spoil" the sound
particularly comparing the adc internal cmos output op amp to today's fast, extended input linearity chips like AD4899 with Gilberts tanh compensated diff pair input built on a dedicated linear semi process
or any of the past decade's precision fet input op amps with 20-60 MHz GBW - some aimed at the opa627637 - or just put out the $ for the opa627
these chips won't have trouble in a properly designed I/V application - should perform far beyond NE5534 and the adc's internal analog circuitry
I assume you have a blocking cap on the input of the amp? Why the 1k series resistor?
I meant 1k resistor between gainclone chip and dacs. It's just reuired input resistor for chip. I'm using AK4396's balanced outputs for instrumentation amplification with LM3875/gainclone chip.
This implementation permits avoiding dc blocking issues (which requiring caps in signal path) if you are not connect ground pin of dac to amp. Just + - pins.
- Status
- Not open for further replies.
- Home
- Source & Line
- Digital Line Level
- AK4395/4396 ultrasonic noise spectrum graphic