Thanks... can I call you Ugly?
I agree that balanced DS can be far less risky to implement than SE, particularly for diyer's, those that on average are likely to have far less experience and skills than designers, perhaps building and/or selling products over decades.
It seems reasonable to conclude that nodes don't actually exist in reality, being a necessity in simulations to predict outcomes often to extreme detail. Yet lines drawn between circuit elements in simulations is not equivalent to physical lines such as grounds, power supply lines or other lines between circuit elements. Lines in simulations between elements are nevertheless points, single nodal point, not lines. So it is that every length of conductor or plane not being simulated (being represented as a point) is subject to suspicion... oftentimes unwittingly neglected. Oftentimes where simulations end, with still a length of wire in hand, prayer begins... and I've been delinquent in my tithing...
🤣 🤣 🤣 I like your sense of humor Hierfi, sometimes I read your post quickly, then don't understand a thing, then come back and in-between words you throw very good humor that's not found by diagonal reading , Nice!
Otherwise, I still like Lo-Fi BSR autodeck, where arm is returned to rest after inner groove is reached. Soon I will make TT for my 8 year old kid from one of those, Ill post it.
Same here, one of my presumptions for this thread was that no head amp is needed as one device can cover more than sufficient gain to bring signal to good level for equalizing.
On power supply, in this thread we are far from discussing that. Front end is not even close to any conclusion.
I also have mixed feelings about PS, on one hand I started to believe in power tools cells, cheap (recently I bought 6 x 2.5Ah 18V power packs at 7 Euro each on discount. 18V is nice number for preamps and we have lot of time to recharge while we sleep, especially since smallest of those are 2.5 Ah. They have integrated battery management to ensure each cell in series is optimally charged. However specific battery pack needs to be studied a bit to check if battery management is influencing discharge (when we use it to power audio) and some sort of battery charge indicator should be inserted visible so we know when to charge. It all sounds merry with little effort.
On the other hand, nice guy on this forum, Grunf, showed me shunt regulators in his development with partner, Very interesting even my brain still doesn't grasp why is it important to have low PS impedance in MHz region for audio. I will take this task in near future...
Nice evening,
Dražen
Well, depends what we do.
If we talk typical hifi preamp we only need double set of contacts for input selection, resistor switch based 4x attenuator (4 gang sliding potentiometers can work but will damage CMRR due their poor tolerances) and 4 bufferes (or 2 fully differential buffers). Phono preamp we discuss here, not much more is needed to go fully differential. Most of DAC's are fully differential in their nature. So far no issue with regular hifi.
If we however go to complex active crossover filters, equilizers or mixing consoles, keeping whole internal signal fully differential will quickly turn into a nightmare, in this cases using differntial recivers and drivers at both end, while keeping signal processing SE makes more sense...
I'm totally impressed that I can talk with you here Hans, not supprised that Bill Whitlock worked with you.
Personally I dropped cable question many years ago when it was hype, had some audioquest pythons or anaconda and slate (anyway some snake and rock names) expensive wires blind tested compared with common power cables (for speakers) and out of box chinch interconects from some cheap vcr... Auditorium seen no difference..... But when you say, I'll revisit the issue.
Hopefully I didn't offend in my previous post Ugly... now that I'm getting much older it seems towards interjecting some creative (oftentimes juvenile and twisted) humour that I seem most wanting to trend. Its lots more fun for me and hopefully more fun for everyone at DiyAudio in not getting too serious over all of this.
I wont be offended. I chose the monicker. Please do carry on with all the juvenile and twisted humor. I would not have it any other way. I do not offend easily.
My apologies for the delayed reply.
My interest is phono circuits of various flavors. When thinking of ways accomplish all the functionality I seek, using a balanced differential topology, there are often instances I can't find reasonable solutions for getting around unacceptable circuit imbalance, usually due to something like the terrible tolerance of commercially avvailable capacitors. I may then feel compelled to convert to using single ended circuit to get the function I want in what I consider to be a more reasonable way. Though it is always a bit of a cringe due to knowing achieving high quality SE circuits is harder.
Sure higher end designs can certainly help sidestep the problems that come with SE circuits, but you definitely have to work for it. The extra work to avoid the risk seems like a pretty convincing reason to stay in balanced differential unless compelled otherwise.
It still seems mysterious what performance benefit is it that would necessarily attract the more talented or experienced designers to convert to SE circuit topologies without some reason like I mentioned could convince me to switch. What is the inhereant advantage? After all why not take the easier , less risky path, especially since as I was aluding to with my DAC/ADC review example, the highest objectively performing devices appear to be incorporating balanced differential circuits.
I often wonder if the apparent audiophile attraction to all things SE is rational. Maybe it is and I don't understand why.
I mentioned the board layout and striving toward more perfect balance, and the layout techniques that could be utilized speculating that perhaps suboptimal design practice may have something to do with large swaths the high end choosing SE in general, Maybe the balanced differential implementations haven't been as good as the could be? Or is there some fundamental limit being encounterred, which SE circuits are able to get around?
I conducted further experiments pushing the limits of the SSM2019 in current mode, in particular to investigate the sonic implications of clipping in the pre-RIAA amplification stage. The conclusion being that the SSM2019 can be operated with higher output levels nearer overload limits, that by doing so can improve sonics and the signal to noise ratios of subsequent stages by feeding higher level signals to those networks. The result sonically was greater reach into the background with corresponding greater dynamic extension and expression, ultimately sounding more real IMO. This was when the gain was set for 2 VRMS / 16KHz bandwidth, being absent of frequency extension recovery by creating a secondary pole in the 2122 Hz network (R2 was set to 0 Ohms below).
The network was tested using only one channel on a breadboard in open air, being absent of shielding or a ground plane under the breadboard. In this environment there was little noticeable random noise when gain was increased to hearing "silent" grooves, with no RFI/EMI interference, despite a bank of fluorescent lights in the vicinity. IMO this has dominantly the result of several factors, using a balanced shielded cable connected per below with OEO (one end only) termination, though the shield was connected to the opposing end of what Hans posted in #302, using the SSM2019 in transconductance mode along with low ohmic return of the +ve and -ve inputs to signal reference ground, and that of boosting the gain. This relies upon the coils being floating and operating at ~ -0.6 Volts.
Reviewing the various termination and shielding networks outlined in "Design of High-Performance Balanced Audio Interfaces" by Bill Whitlock their doesn't seem a more formidable and simplistic means to interface the galvanically isolated coils in the handling of variant RFI and EMI type noise than per shown below, seemingly requiring no RF input filtering other than perhaps the need of some RF beads being used. The symmetry is perfect... at least on paper.
The network was tested using only one channel on a breadboard in open air, being absent of shielding or a ground plane under the breadboard. In this environment there was little noticeable random noise when gain was increased to hearing "silent" grooves, with no RFI/EMI interference, despite a bank of fluorescent lights in the vicinity. IMO this has dominantly the result of several factors, using a balanced shielded cable connected per below with OEO (one end only) termination, though the shield was connected to the opposing end of what Hans posted in #302, using the SSM2019 in transconductance mode along with low ohmic return of the +ve and -ve inputs to signal reference ground, and that of boosting the gain. This relies upon the coils being floating and operating at ~ -0.6 Volts.
Reviewing the various termination and shielding networks outlined in "Design of High-Performance Balanced Audio Interfaces" by Bill Whitlock their doesn't seem a more formidable and simplistic means to interface the galvanically isolated coils in the handling of variant RFI and EMI type noise than per shown below, seemingly requiring no RF input filtering other than perhaps the need of some RF beads being used. The symmetry is perfect... at least on paper.
I just came across the expensive $22,500,- Audio Research phono amp, the Reference Phono 3.
What quite surprised me was it’s output impedance, 200R in SE and 400R Balanced 🤔
Hans
What quite surprised me was it’s output impedance, 200R in SE and 400R Balanced 🤔
Hans
My demon 😅 (Denon DN-A7100) that is used in our "cinema room" has exactly 200r termination at xlr outputs after 5532. It is made to run tens of meters of cables for surround. Seen it in different pro gear too, but not over 200ohm I admit, 50 to 100R is much more common.
Particularly since the RIAA and output drive is via 6H30P dual triodes. There are 6 in the audio (3 per channel), and one as a driver in the HV PSU.
The price of them seems to be ridiculous. Hundreds of dollars for a pair.
But I don't know why 200 ohms in series with each output. Perhaps to prevent the ridiculous priced valves/tubes from blowing up if the output is inadvertently shorted.
Craig
Hi,
In meantime I started testing some motors for TT here: https://www.diyaudio.com/community/...eded-with-motor-and-drive.412984/post-7789788
While on holidays, reading and contemplating, I started to love SSM2017(9) diagram more and more. It allows for CFB, huge gain, decently low noise and so on.
However we cant aces all options as there are only 8 pins available. Contemplation goes in direction to make semi discrete clone, without output integrator as it is not needed for fully differential.
What do you say about following; being just concept sketch?:
Cart can be placed instead of Rg to make it trans impedance amp too, as SSM2017 can do...
In meantime I started testing some motors for TT here: https://www.diyaudio.com/community/...eded-with-motor-and-drive.412984/post-7789788
While on holidays, reading and contemplating, I started to love SSM2017(9) diagram more and more. It allows for CFB, huge gain, decently low noise and so on.
However we cant aces all options as there are only 8 pins available. Contemplation goes in direction to make semi discrete clone, without output integrator as it is not needed for fully differential.
What do you say about following; being just concept sketch?:
Cart can be placed instead of Rg to make it trans impedance amp too, as SSM2017 can do...
For voltage mode output the simplistic means to do differential out seems just to use 2 SSM's per below... seems likely to reduce noise as well. Keep in mind that feedback is applied to the gain setting resistors independently as internally not to the input terminals.
Last edited:
I simulated just a possible version of the circuit as a transimpedance amp with Rcart=10R and diff. gain of 46dB.
The two Opamps where ordinary 5nV/rtHz versions but the transistors the very low noise ZTX851 with 5mA feeding them.
THD with 10Khz@10mV and the gain of 46dB was -135dB, while noise from 20Hz to 20Khz was 74nV or 520pV/rtHz, which is all very good.
Hans
Superb, thanks Hans!
When you are at it, can you change input to 0.35mV , and later Rcart to 40 ohm for comparison, so we see both ends, low and high R MC?
PS, gain will drop then, buit that's how it is...
I need much more info.
What do you want to see with 0.35mV.
I can change the Cart to 40R, but keep the gain at 46dB, or ?
Hans
Hi Hierfi,
I can confirm that circuit you draw is very good, excellent actually!. It is qualified by more than 20 years of use 🤣
That is exactly my original preamp posted at beginning of this thread https://www.diyaudio.com/community/...ono-preamplifier-thoughts.414816/post-7731430
However good they are, SSm2017-9 are tuned as mic preamp and microphones normally have much higher Z than MC carts . By rconstructing the model we can fine tune it with better input transistors and higher bias currents as Hans just did 🙂
Yes please , than you need to change 2x Rf too (R2 and R3).
It seems that you used 10mV as input signal? In my measurements using higher power signal produced much better results than when using "standard" 0.35mV of common MC cart. Just want to see realistic case..