Each shift register is slightly non-linear because the high output resistance is different from the low output resistance. Looking at both shift registers together, the number of high outputs is independent of the data. Hence, the effect of the unequal resistances largely cancels when you look at the differential output.
Regarding matching, when you look carefully how the flip-flops are connected, you will see that there are really four four-tap shift registers, two for the non-inverted and two for the inverted data. U14 and U16 each form a shift register for the non-inverted as well as the inverted data, but the flip-flops that handle the inverted data in U14 handle the non-inverted data in U16 and the other way around. That's all done for matching: taking advantage of the close matching between flip-flops on the same chip and cancelling the effect of unequal on-chip wiring to the supply and ground pins.
Regarding matching, when you look carefully how the flip-flops are connected, you will see that there are really four four-tap shift registers, two for the non-inverted and two for the inverted data. U14 and U16 each form a shift register for the non-inverted as well as the inverted data, but the flip-flops that handle the inverted data in U14 handle the non-inverted data in U16 and the other way around. That's all done for matching: taking advantage of the close matching between flip-flops on the same chip and cancelling the effect of unequal on-chip wiring to the supply and ground pins.
Last edited:
No need to turn this into Marcel's dac vs. Andrea's dac. I'm not aware of any impartial reviews of Andrea's dacs and have not seen any schematics or measurements so comparison with Marcel's dac is only based on your claims. While you may like Andrea's dac it is only your subjective opinion based on uncontrolled listening evaluation. And not even an impartial opinion since you have invested quite a lot on Andrea's dac and you seem to be socially acquainted with him.
@Mark, looking to me more like euphonics of your transformer based line Amp ( and your SE power Amp) that has hit a sweet spot with you regardless of what DAC you use. And you are doing a lot of stuff to tune this passive filter to match accordingly. Probably, others will not be able to replicate this arrangement. All the same, just carry on with further improvements on the input side as you were planning, thanks for your efforts
@Marcel, after checking with BenchMark, it appears that their AHB2 power amps that I am using may not match well with your DAC output level (2V rms BAL). AHB2 has fully differential inputs(ok) but requires 9V rms drive at 0dBFs. Both my AKM and EES type have this high output drive. It is possible to set gain to 2V inputs but less than optimal. Also, they do not recommend SE drive (via adapters or transformers) other than for testing or compatibility with field equipment.
Any chance of doing 9V or close enough output but optimally, just to keep the playing fields level?
Thank you
Any chance of doing 9V or close enough output but optimally, just to keep the playing fields level?
Thank you
Acko, this is from their website:
From what I gather, it`s mostly to ensure noise ratios are met:
The SENSITIVITY switch must be set properly in order to maximize the signal to noise ratio of your audio system. If this switch is set too low, you will not be able to drive the amplifier to its rated output. If this switch is set too high, the noise produced by your preamplifier will be amplified more than is necessary. If you are using an unbalanced input, you must use one of the top two switch positions. If you are using a balanced input, you will usually need to use one of the bottom two switch positions. Set this switch all the way down if you are using professional balanced sources. Raise the switch to increase the output level. Lower the switch to decrease the output level. When set properly, you should be able to drive the amplifier to the onset of clipping when your volume control is set near maximum
Input Sensitivity
- Low-Gain = 22 dBu (9.8 Vrms), Gain = 9.2 dB
- Mid-Gain = 14.2 dBu (4 Vrms), Gain = 17.0 dB
- High-Gain = 8.2 dBu (2 Vrms), Gain = 23 dB
- Use Mid-Gain or High-Gain settings for unbalanced inputs
- Unbalanced inputs require RCA to XLRM adapter cables
From what I gather, it`s mostly to ensure noise ratios are met:
The SENSITIVITY switch must be set properly in order to maximize the signal to noise ratio of your audio system. If this switch is set too low, you will not be able to drive the amplifier to its rated output. If this switch is set too high, the noise produced by your preamplifier will be amplified more than is necessary. If you are using an unbalanced input, you must use one of the top two switch positions. If you are using a balanced input, you will usually need to use one of the bottom two switch positions. Set this switch all the way down if you are using professional balanced sources. Raise the switch to increase the output level. Lower the switch to decrease the output level. When set properly, you should be able to drive the amplifier to the onset of clipping when your volume control is set near maximum
Last edited:
So just the good ol' "modding", "matching" "op am rolling" kinda activity. All to the tunes of studio recordings.... which no one know how they sounded at time of mastering. But the confidence and wording is mesmerising and we all tend to believe that there is something special going on - which there isnt of course.
Stay cool guys. Continue to do the science - it will succeed while dabbling is bound to get stuck. If it isn't already. I think it is.
//
Ok, thanks… I will try the 4V/2V rms sensitivity and see,
Also, btw although the inputs are balanced the outputs are SE. So like many other amps there is conversion from differential internally to SE outputs. There is also a bridged mono mode so BAL to BAL possible.
Last edited:
Absolutely. We use differential pairs for all sorts of things.
Last edited:
That gain is at the 2nd stage. IIRC, the best practice is to change the gain at the first stage itself. Something to do SNR. That’s why the same way not really the best to do it at the Amp side although the flexibility is there for the AHB2 amp. Factory default is 9V rms input. AKM tuned to 9V in their default also. ESS used to be 4V rms but with their new 9039PRO now tuned to 9V right at their new IV stage,
Regarding the comparing of dacs. To clarify a bit, Acko has been asking about both dacs in other than just this thread. Also, the dacs are both RTZ FIRDACs so it is kind of natural to compare them. They do differ in design philosophy in some ways however, with one more like leaving nothing to chance, no stone unturned, and so it costs more. The other is going more for elegance and aiming to only do as much as really needs to be done to get the desired result, so it is less costly. Some people may or may not want to know about these things. If and Acko and or Marcel feel its a problem and or ask me to restrict discussion to Marcel's dac only, I am willing to do that. If not, I will continue as I see fit.
Regarding the transformers that are engineering prototypes. Right now they are NDA. I keep hoping the day will come sooner than later that I can say more. Not up to me though, not my transformers, probably not the final version, etc.
Regarding vague comparisons of output stages, in all cases Marcel's, Andrea's, and the transformers, those things wouldn't be out there if the people who designed them didn't think they were good enough. In all cases the stages have been listened to and people have expressed satisfaction about the sound in their systems. This is despite the fact that all the output stages sound different. As was mentioned recently by another member, lots of parts in systems including amplifiers, speakers, etc., to some extent have their own sound. FR, HD/THD and noise floor type measurements do not completely capture those differences. I can only try express my own opinions on these things and hope that will be good enough for what one person can do. If people want to know more details than what I write here they can always PM and ask what they want. Don't know what the responses might be until I see what the individual questions are.
Regarding the transformers that are engineering prototypes. Right now they are NDA. I keep hoping the day will come sooner than later that I can say more. Not up to me though, not my transformers, probably not the final version, etc.
Regarding vague comparisons of output stages, in all cases Marcel's, Andrea's, and the transformers, those things wouldn't be out there if the people who designed them didn't think they were good enough. In all cases the stages have been listened to and people have expressed satisfaction about the sound in their systems. This is despite the fact that all the output stages sound different. As was mentioned recently by another member, lots of parts in systems including amplifiers, speakers, etc., to some extent have their own sound. FR, HD/THD and noise floor type measurements do not completely capture those differences. I can only try express my own opinions on these things and hope that will be good enough for what one person can do. If people want to know more details than what I write here they can always PM and ask what they want. Don't know what the responses might be until I see what the individual questions are.
Possibly, although such general rules should be tested in practice. Here is another comment on this.
As you don't have the schematics of Andrea's dacs it is somewhat difficult to judge whether or not nothing has been left to chance. E.g. the reclocker & isolator you presented in #1259 seems to use single flip flops which may lead to metastability.
Didn't say I haven't seen any schematics at all, only said for this DSD dac. Also didn't say I don't know how to reverse engineer a dac from inspection, scope measurements, etc.
One concern I have here is that its not my place to give away certain details of Andrea's technology no matter how I come by the knowledge. That's up to him. Or someone else, if they want to buy one to reverse engineer.
Regarding the reclocker board I posted a pic of, think I said its an accessory board for the FIFO board. The FIFO board transmits signals synchronously so that metastability at the accessory board should never occur. It it does then there is a bad cable or something that needs troubleshooting.
More strictly speaking, it cannot be literally true that 'nothing can be left to chance' by an engineering design made in this physical universe. To "leave nothing to chance" is an expression that most people know the meaning of. Same can be said for 'leave no stone unturned.' Andrea's dac contains no literal stones.
One concern I have here is that its not my place to give away certain details of Andrea's technology no matter how I come by the knowledge. That's up to him. Or someone else, if they want to buy one to reverse engineer.
Regarding the reclocker board I posted a pic of, think I said its an accessory board for the FIFO board. The FIFO board transmits signals synchronously so that metastability at the accessory board should never occur. It it does then there is a bad cable or something that needs troubleshooting.
More strictly speaking, it cannot be literally true that 'nothing can be left to chance' by an engineering design made in this physical universe. To "leave nothing to chance" is an expression that most people know the meaning of. Same can be said for 'leave no stone unturned.' Andrea's dac contains no literal stones.
Last edited: