Hi zjazd,
I have a test unit here from the last run of boards if you're interested. Send me a PM.
mcluxun:
Looking good!
Great layout, and great choice of power supplies. Dual Mono is probably overkill, but if you want the best of the best you've got a pretty killer build there!
Cheers,
Owen
Hi Owen,
Not wanting to be a d!ck to mcluxun, but ...
I remember looking at the diyinhk layout for those tps7a4700 regs a while back and thinking they missed a few tricks with the layout, especially the sense pin I wouldn't expect them to get the datasheet value for noise they proudly write on the pcb silkscreen, but that's just a gripe of mine where people assume the chip under any conditions will work to maximum datasheet spec.
Should work well for this application but ... I honestly don't see them being a notable improvement on Owen's PSU v2. Should be roughly equivalent on paper I'd have thought.
If I was going for something overkill/tweaking in PSU department for this amp, a battery feeding Owen's PSUv2 is where I would go.
My 2c.
Cheers,
Chris
Input pot for the BAL-BAL
Hi All,
Maybe somebody can help me with my next BAL-BAL.
I want to include the BAL-BAL in my DDDAC taking the signal directly from the balanced output of the Sowter trafos.
My problem is relevant to the dedicated pot that I want to install between the Sowters and the BAL-BAL.
I look in Hificollective and they have only 100K as balanced and I don't know if it is the correct value.
Any suggestion?
Thanks and Regards,
Enrico
Hi All,
Maybe somebody can help me with my next BAL-BAL.
I want to include the BAL-BAL in my DDDAC taking the signal directly from the balanced output of the Sowter trafos.
My problem is relevant to the dedicated pot that I want to install between the Sowters and the BAL-BAL.
I look in Hificollective and they have only 100K as balanced and I don't know if it is the correct value.
Any suggestion?
Thanks and Regards,
Enrico
Also, the line regulation of those chips looks to be nothing special. The load regulation neither? And the noise is a higher at 15Volts than 5Volts, 13uV vs. 4.17uV
Not a Salas or Jung/Didden super reg? Which it looks like having lower noise than batteries, especially as current rises, and perhaps lower output impedance, load regulation?
I've been trying to get to the bottom of this in recent weeks so I can build 'the best' regulator for my Bal-Bal, and other items. I already improved matters with the addition of one 100uF 25V Black Gates across each of the four Nichicon Polymer caps. The BG's did their normal thing of more transient response and louder bass; demonstrating that the three transistor discreet regulators I have have poor load regulation? They don't have separate load sensing wires either and years ago, with the DAC and output stage that used to feed, I found the wire I used to convey the ± rails and earth had a large effect on the sound quality and I eventually used 0.71mm sold silver. I've been ignoring the quality shortfalls for many years but now I seem to have the will to attempt to address this issue. I've not tired different PSU wire with the Bal-Bal, or anything apart from the BG's, but that they had so much effect suggests there is more quality available to the Bal-Bal that I currently getting. I'd like to try load sensing wires right on the output chip ± pins themselves.
Here's one bit I surfed up about batteries vs. regulators: Simple Voltage Regulators Part 1: Noise Batteries at the bottom of the page.
Compared to these TPS regulators, and some aspects of batteries, it looks like the line, load, and noise are all better with some Jung super reg derivatives, and the Salas is perhaps 10dB lower in noise again.
Last edited:
Is line regulation terribly important for this application?
Can you let me know what headphones you're using? We can try running some numbers for the sorts of load we're expecting from the headphone amp.
I'm no-one's idea of a subjectivist so I might not be able to give you the info you're after perhaps? I'll try to help though.
If we look at Owen's post here - http://www.diyaudio.com/forums/vend...ble-here-bal-bal-se-se-lpuhp.html#post3699950 (post #144) we can see the noise profile of the PSUv2 and the earlier version of the PSU with LM3xx regulators. The TPS7A3301 that Owen uses there is looking pretty good. My thoughts were that a NiMH battery pack with a trickle charge circuit would clean up the mains related components and with reasonable capacitance and these regs we'd be looking at some pretty solid performance, especially considering the application has a reasonable PSRR to start with! The opamp PSRR is better than -100dB throughout the audio band.
Look a bit further down that page at the 1kHz plot for the BAL-BAL - we only see a small amount of 60Hz noise in the noise floor and the rest is headphone amp + DAC and ADC distortion, it'd be a VERY intense exercise to identify exactly what distortion contribution the BAL-BAL had to that measurement with a reasonable level of confidence.
I've got a PCB here that I designed with rectifier + snubber (Qusimodo style), LM317 pre-regulator, TPS7A4700 at output and 4 wire output. I am planning on doing some measurements of its performance at some point too to see if the extra bits help the mains related noise components. I've had those PCBs sitting here for some time though and need to build another one to test (I sent some away to a friend to test and didn't build one for myself at the time) then I need to build a LNA to complete the testing of noise floor etc. That'll be a while before I get around to that, there are higher priorities for my time at the moment. The though honestly never crossed my mind to use this for the BAL-BAL though, I don't see the need in this application.
If you're really not convinced to sit back and enjoy listening to a great headphone amp and not worry about the tweaking then - some further reading that might be worthwhile is the comparison of a bunch of regulators that jackinnj did in Linear Audio 'zine Vol 4 the article is well worth the price of admission for the electronic version or the hardcopy. There are some charts on the downloads page under the heading "Color graphs for Jack Walton's regulator article in Vol 4"
Cheers,
Chris
I don't personally know. It was with my previous DAC as every change made before the regulators had an obvious effect on the sound. I presume if the line regulation had been perfect, no change would have been heard.
HD800 with hard wired solid silver leads. I don't know if that the best wire but it's better sounding than the stock Sennheiser stuff; clearer, transient response, the stock leads were comparability dull, soft, muddled, blurred.
Well, I guess that's what I've been doing for the last couple of years, but I was aware that my sound was not as good as some aspects of other systems I've heard. I've been going to a good sounding cinema a lot in recent months and I'd like that bass clarity and dynamics from my headphones. And I have Anthony Gallo Strada speakers and that system produces a faster, punchier kick drum than my headphones, fed from the same source; a modified Auzentech X-Meridian soundcard with a load of polymer caps and others replacing the originals, DC coupled, LME49990 op-amps, output stage direct silver wired to XLR's to the Bal-Bal. I'm sure it could be better, but it's not too terrible. Using Foobar 2000. A pipe dream is to take I²S from the CMedia DSP chip to a DAC and output stage mounted outside the case fed from a much better PSU.
I like the CMedia DSP headphone sound sound effect for TV and music. For music it's nearly the same as sitting in front of a big pair a very high quality monitors, for theatre it does a pretty good surround, not quote as distinct at the rear as my old surround speaker system, but a lot better than without any 'cross feed'. I couldn't tolerate music without cross feed, I'd have to go away from headphones and try and get a dedicated soundproofed room for speakers again. I'm not affluent enough to give the Smyth Research Realiser a try. If there's an alternative to either of these I'd like to know about it
To run the Gallo speakers I unplug the Bal-Bal and plug in the silver leads which are hard wired to the monoblock amps.
Ah, thanks, I was looking for that
I'll look again in a while.I've been following that thread and was wondering about forking out the $4 or so for that article. I was just reading about another, or the same, 'Regulator Bake off' with more of those graphs and other info. I started doing more reading a week or two ago but there is so much to sift through for the gems and alas I'm a very slow reader (Meares Irlen syndrome). I would have been using a super teddy reg for the DAC a few years ago if those FETs(?) hadn't gone out of production, then I lost interest for a couple of years.
Cheers, Ian
Last edited:
Just in case it helps longer term reliability, I stuck these 87°C/W heat sinks on the OPA1632's: Fischer Elektronik ICK SMD A 8 SA
ICK SMD A 8 SA - FISCHER ELEKTRONIK - HEAT SINK, FOR SMD, 87°C/W | Farnell element14
There were a slightly tight fit as are the next size up and overlap adjacent parts.
Used thermal adhesive tape:
T Global LI-98-150-025
Thickness: 0.25mm
0.8W/m.K Thermal conductivity.
Thermal Impedance: 1.8°C/W
ICK SMD A 8 SA - FISCHER ELEKTRONIK - HEAT SINK, FOR SMD, 87°C/W | Farnell element14
There were a slightly tight fit as are the next size up and overlap adjacent parts.
Used thermal adhesive tape:
T Global LI-98-150-025
Thickness: 0.25mm
0.8W/m.K Thermal conductivity.
Thermal Impedance: 1.8°C/W
Hi,
Seems getting RF in via the headphone wires acting as an antenna. As this is a very wideband amp and fast slewing as well, it likely is pretty susceptible to this.
As others have suggested adding a resistor in series with the headphones could help but would lower the damping factor quite a bit.
What I would do instead is add a beaded lead. A wire with a RF filter bead on it - in series with the output. Depending on the size and frequency range of the beaded lead, might have to add two or three in series to get enough inductance to stop the RF from coming into the amp. Would have very little effect on the damping factor at audio frequencies. FWIW.
Cheap source: allelectronics.com
Regards,
Greg
Seems getting RF in via the headphone wires acting as an antenna. As this is a very wideband amp and fast slewing as well, it likely is pretty susceptible to this.
As others have suggested adding a resistor in series with the headphones could help but would lower the damping factor quite a bit.
What I would do instead is add a beaded lead. A wire with a RF filter bead on it - in series with the output. Depending on the size and frequency range of the beaded lead, might have to add two or three in series to get enough inductance to stop the RF from coming into the amp. Would have very little effect on the damping factor at audio frequencies. FWIW.
Cheap source: allelectronics.com
Regards,
Greg
more on eliminating RFI on OpAmp output pin
Filtering needs VoltageDivision action. Hence 2 impedances are needed.
The FerriteBead provides one. The 2nd might be some small or moderate
capacitance added to OpAmp output pin.
Is 100pF OK? The OpAmp output transistors likely have 10/20/30pF of
implanted junction capacitance, on chip. Those interact with the Bead to
attenuate the RFI. More capacitance will help, unless the OpAmp gets
squirrely with lumped Cload.
tankcircuitnoise
Filtering needs VoltageDivision action. Hence 2 impedances are needed.
The FerriteBead provides one. The 2nd might be some small or moderate
capacitance added to OpAmp output pin.
Is 100pF OK? The OpAmp output transistors likely have 10/20/30pF of
implanted junction capacitance, on chip. Those interact with the Bead to
attenuate the RFI. More capacitance will help, unless the OpAmp gets
squirrely with lumped Cload.
tankcircuitnoise
Have you guys ever measured the distortion of a ferrite bead when you start passing reasonable current through it in series applications?
It's bad... it's very bad. It will take the 0.0001% THD of this amp and turn it into between 1% and 0.01% depending on the load, drive level, and size of ferrite bead used.
Ferrite beads have a very strong non-linearity caused by magnetic hysteresis. I've worked on this in-depth trying to pass EMI/EMC requirements for headset jacks on mobile devices. Passing CE "Audio Breakthrough Tests" can require large 600 ohm ferrites which resulted in roughly 1% THD into a 16 ohm load.
Bottom line:
You'll do more harm than good trying to solve a problem you don't even know you have yet. That's not good engineering.
Use proper shielding, good grounding, proper system gain planning and mind your input and output impedance.
There are a dozen other other ways to address this concern, and ferrite beads should be an absolute last resort.
Regards,
Owen
It's bad... it's very bad. It will take the 0.0001% THD of this amp and turn it into between 1% and 0.01% depending on the load, drive level, and size of ferrite bead used.
Ferrite beads have a very strong non-linearity caused by magnetic hysteresis. I've worked on this in-depth trying to pass EMI/EMC requirements for headset jacks on mobile devices. Passing CE "Audio Breakthrough Tests" can require large 600 ohm ferrites which resulted in roughly 1% THD into a 16 ohm load.
Bottom line:
You'll do more harm than good trying to solve a problem you don't even know you have yet. That's not good engineering.
Use proper shielding, good grounding, proper system gain planning and mind your input and output impedance.
There are a dozen other other ways to address this concern, and ferrite beads should be an absolute last resort.
Regards,
Owen
Not that I've seen... it's primarily related to the size of the part and the impedance. Smaller 0201 10-22ohm ferrites exhibit this problem to a much smaller degree, but they are also less effective than larger values for reducing RF noise. Larger 0603 and 0805 parts from 600-1000 ohms are very problematic.
Ceramic and air core parts do not seem to exhibit this behaviour, but they too are less effective at reducing RF coupling as they are generally high Q and need to be tuned for a specific problem frequency.
Regards,
Owen
If I understand the use of ferrites correctly, then they must be part of a two component filter, just as tank says
A ferrite around BOTH signal wires (both the signal currents cancel through the ferrite and thus the ferrite adds no impedance to the signal route) followed by the capacitors to the chassis to take the majority of interference to "earth".
The filter is arranged to attenuate the interference and NOT attenuate the wanted signal. This requires the high impedance series element to be in the feed facing the interference and the low impedance element on the "quiet" side of the filter.
A ferrite around BOTH signal wires (both the signal currents cancel through the ferrite and thus the ferrite adds no impedance to the signal route) followed by the capacitors to the chassis to take the majority of interference to "earth".
rfi chokes and clean VDD
A "large" distortion-free inductor can be wound around a 1/4 watt resistor.
I assume the resistor's body provides no distortion under magnetic field.
Another topic I dig into is ---- ringing of VDD, triggered by OpAmp surge
currents.
Of course incoming RFI, arriving via the headphone wires, will find a place to go as that high frequency explores all possible paths. Thus the VDD network (Caps, inductors, resistors, beads) can be *designed* to address both the surge current ringing and the RFI_ringing.
Summary, for now: I view the VDD network to also be a design issue.
tankcircuitnoise
A "large" distortion-free inductor can be wound around a 1/4 watt resistor.
I assume the resistor's body provides no distortion under magnetic field.
Another topic I dig into is ---- ringing of VDD, triggered by OpAmp surge
currents.
Of course incoming RFI, arriving via the headphone wires, will find a place to go as that high frequency explores all possible paths. Thus the VDD network (Caps, inductors, resistors, beads) can be *designed* to address both the surge current ringing and the RFI_ringing.
Summary, for now: I view the VDD network to also be a design issue.
tankcircuitnoise