beats me why but God be my witness, DAC ON -> no more TV (actually only a number of channels were affected). DAC OFF -> TV back.
I'd take a wild guess and say that everything was oscillating like crazy.
but, there are discrete regs. and a 1000V/us current feedback amp.
yep, oscillating and broadcasting lol the bipolar junctions on the chip were likely rectifying the RF before passing it on as a signal and not just over cables either, as you found out. then it was probably being passed back as RF and electrical feedback to the wide bandwidth discrete regulator and around we go again!!
Oh well, I guess its the tpa6120/tpa6012 you are talking about on the output of a Ti dac, there is an app note to that effect, but it has a filter. bitch of a chip to get right and there are plenty of examples of doing it wrong. its originally an ADSL driver as the 6012, same chip under a different name.
sux that you had to change amps before you realized exactly where the problem was.
Oh well, I guess its the tpa6120/tpa6012 you are talking about on the output of a Ti dac, there is an app note to that effect, but it has a filter. bitch of a chip to get right and there are plenty of examples of doing it wrong. its originally an ADSL driver as the 6012, same chip under a different name.
cool! where can I get one? its my first DIY project and this sounds ideal...
sux that you had to change amps before you realized exactly where the problem was.
Last edited:
I am no longer surprised when commercial items are incompetently designed and/or incompetently built. EMC issues are routinely ignored or deliberately subverted. Part of the blame must rest with modern EE education; part with modern fashion so marketing buttons take precedence over function; part with cheap overseas manufacture.
sorry yes THS6012, not tpa, its the same chip, but with a couple of package variations available.
yes, well, when we can often be labelled as paranoid nit-picking Nazis here on this technical forum (sometimes I wonder) for mentioning EMC issues, without any peer review i'm sure all sorts of crap happens. Some people think that if they cannot hear it its not happening and some may well hear that crispy RF as added detail, meanwhile claiming to be able to hear things that may only have a minute effect up in bat territory, so its not surprising some make it to sale.
That doesnt make it any less aggravating for Mr PP though.
yes, well, when we can often be labelled as paranoid nit-picking Nazis here on this technical forum (sometimes I wonder) for mentioning EMC issues, without any peer review i'm sure all sorts of crap happens. Some people think that if they cannot hear it its not happening and some may well hear that crispy RF as added detail, meanwhile claiming to be able to hear things that may only have a minute effect up in bat territory, so its not surprising some make it to sale.
That doesnt make it any less aggravating for Mr PP though.
Last edited:
I too feel that EMC is one of the hardest things to get right.
even harder to understand why one chooses the fastest amp available for the buffering task.
oh yeah i'm totally aware of that issue and ive been through it all before, some disastrous mods to headphone amps have been made. that being said, follow the app note and you get a pretty crappy headphone amp, because the output resistor is outside the loop the low impedance drive is terrible and headphones are designed with lower and lower impedance these days, so I think the app note itself is a bit dubious.
Its actually quite good with higher Z loads and jcx came up with a reasonable design that included it in the loop, though even he had some initial layout bugs as I remember.
God knows why you would use it for a regular dac output buffer though, did it have the opa2134 as well or feed the buffer directly? the filter was around the OPA from memory, so maybe he got rid of that whole section?
Last edited:
qusp, the info about the 6120 is useful. just found the THS6120 datasheet which seems to be ehxaustive.
the buffer is fed directly. I'll look up jck's design tomorrow, thanks. WM8741 is at the left, you can see the traces going directly to the 6120.
the buffer is fed directly. I'll look up jck's design tomorrow, thanks. WM8741 is at the left, you can see the traces going directly to the 6120.
[sarcasm]filter? what filter? caps ruin sound. hiss and oscillations sound good.[/sarcasm]
Attachments
Last edited:
audiophile approved, distant, leaded film decoupling caps instead of those crappy little ceramic ones crowding the chip too!! I want it even more!
oh yeah its not a bad app note and pretty thorough, as long as you dont expect it to drive modern headphones with any bass. they do go into a fair amount of detail, I just think whoever the designer was from the past when headphones were 300-600Ω and above, hes not alone there, quite a few commercial app notes include output resistors outside the loop.
actually I just did a quick google to see if I could link it, jcx had his amp design on his site, but also on head-fi.org and he actually included the entire tpa6120 in side the FB loop, so was a full multiloop design. the tpa has highish input bias current which this helps to correct as well. It can otherwise produce an uncomfortable amount of DC offset because of this
couldnt find it in the time I have available, perhaps you could PM him here on the forum. he has used this chip a LOT and comments in many threads about it, so with any luck he may have already read this post.
With capacitive loads/cables and a 2.2R resistor you could be getting ringing/oscillation there too. Looking at that cheap 2 layer layout, hes criss crossed the FB loop all over the place, even under several small SMD resistors, that by itself could cause issues with a chip this fast
oh yeah its not a bad app note and pretty thorough, as long as you dont expect it to drive modern headphones with any bass. they do go into a fair amount of detail, I just think whoever the designer was from the past when headphones were 300-600Ω and above, hes not alone there, quite a few commercial app notes include output resistors outside the loop.
actually I just did a quick google to see if I could link it, jcx had his amp design on his site, but also on head-fi.org and he actually included the entire tpa6120 in side the FB loop, so was a full multiloop design. the tpa has highish input bias current which this helps to correct as well. It can otherwise produce an uncomfortable amount of DC offset because of this
couldnt find it in the time I have available, perhaps you could PM him here on the forum. he has used this chip a LOT and comments in many threads about it, so with any luck he may have already read this post.
With capacitive loads/cables and a 2.2R resistor you could be getting ringing/oscillation there too. Looking at that cheap 2 layer layout, hes criss crossed the FB loop all over the place, even under several small SMD resistors, that by itself could cause issues with a chip this fast
My question is how does the poster *know* that his measurement methods and equipment actually was *measuring* what he thinks he was *measuring*? Especially if he doesn't know where to buy small quantities of 2-layer PCBs (trivial, they are all over the place), nor can afford what he calls "a very good scope".
Dunno? Maybe the thing he got sucks, but it sort of makes sense to qualify the complaint(s) - also to be certain that there is not something awry (inadvertently perhaps) in how the unit is being used, or how it is being tested.
The other thing that occurred to me, is contact the mfr first. There may have been one or more errors on the assembly line for a given run, that did not get caught. For example, that resistor might have ought to have been a 22 ohm, not a 2.2 ohm. Etc...
Dunno? Maybe the thing he got sucks, but it sort of makes sense to qualify the complaint(s) - also to be certain that there is not something awry (inadvertently perhaps) in how the unit is being used, or how it is being tested.
The other thing that occurred to me, is contact the mfr first. There may have been one or more errors on the assembly line for a given run, that did not get caught. For example, that resistor might have ought to have been a 22 ohm, not a 2.2 ohm. Etc...
Last edited:
not for sale!
some of jcx's posts showed up while googling.
you were right about DC offset, I measured ~0.2V. I guess I'm lucky that I'm using it with a cap-coupled amp (fortunately the ones making the amp have some sense).
oh, and to make it clear to everyone: the 6120 is used both as headphone amp and line driver in this unit.
of course. especially when the selected part is not exactly the typical line driver. and when it's 75% more expensive than a LM4562.
you know what I think, bear? that you once were or still are in the manufacturing business and you got tired of complaining customers. I can totally understand that. yes, some can't tell + from - and blame it on the manufacturer. but some are right.
yes, I did not know that there are companies that make 2-sided PCBs in very small quantities. maybe I don't care. maybe that's why I pay manufacturers the typical profit margin of 200%-400%, because I'd rather have them do the dirty work. yes, because the way I see it, the profit margin is the price I pay for competence, know-how, R&D, Q&A, quality testing/measurement gear etc. at least the competence and Q&A parts seem non existent in this case.
looks to me like qusp is the only one who's getting my point. he even helped by giving some more details about the output driver. I thank him for that.
I checked the specs of my amp and they list a SNR of 95dB. guess what? with inputs shorted there is barely any noise, detectable only ear against tweeter. but when I power the DAC up the noise is audible from half meter away. and this with a 70W in 4 ohms amp and 83dB W/m speakers. the out level of the DAC is around 1.5 volts RMS and about 2V RMS gives full rated amp output.
to detect the hiss I used a very trusty measurement system called my two ears. I trust them.
since you mentioned it, why don't I give you the full story?
when I first received the DAC and powered it I was shocked, the hiss produced at normal levels was audible from outside of the room and I'm not exaggerating a bit. obviously, the unit wasn't even powered by the manufacturer, and trust me, it's not made in quantities of thousands. I sent it back and it was solved. and then one day I connected to it a power amp (no volume control), intending to use the internal volume control of the DAC. shock and horror again, the hiss was back, audible and annoying from the listening position. turned out that it was always there, but at regular listening levels it got attenuated by the amp's volume control. well, I never bothered to check before with full amp volume. what I measured then was around 60 dB SNR. I sent it back to the manufacturer and this is what I have now: the hiss is not audible from listening position anymore but it exceeds the one made by the amp by a few times.
and carelessness could explain the 2.2 ohm resistor. because in the course of my investigations it turned out that the XLR connectors were mounted upside down, so pin 2 went to GND instead of pin 1. I never used the balanced outs but such things are clear signs. this with a product having "manufactured under strict quality control" written at the beginning of the user's manual.
funnies thing of all is that after all this manufacturer tells me "if you're still not satisfied you can always upgrade to one of out better products". yeah, the ones that reached a state-of-the-art SNR of 90dB. at least he has a sense of humor.
and, again: I'm not even complaining, I have no intention to ask for a refund. I'm just ascertaining a state of matters, and that is the buzzword qualification as qusp calls it, ESPECIALLY when it's not backed up by actual engineering. I see more and more reviews that first post pics of the gear with the lid off, commenting on the "hefty toroidal" etc. yeah, right, so it must be good.
Last edited:
whenever I buy used, I do some testing first, especially with amps. I once burned a EUR 100 woofer with an used amp that had the output devices replaced with whatever transistors were at hand and had identical cases.
I guess I'll start doing the same with signal sources.
forgot to mention that I contacted two owners of the same product which had the same hiss issues.
true, the unit is being used in the most awkward way possible: connected to an amp in order to play music. problem persisted with 2 amps of very different designs (one class D, one A/B, different manufacturers, very different design philosophies etc). tried it in 2 different rooms, in order to eliminate the possibility that it's noise pickup. and 2 other owners had identical issues.
yes. idiot user, obviously.
some good info about 6120 implementation:
Sjöström Audio - QRV07 Headphone amp
looks like peranders knows what time it is.
Sjöström Audio - QRV07 Headphone amp
looks like peranders knows what time it is.
looks like Bruno Putzeys also knows what time it is:
(this is about high-end pricing but touches some some aspects we're dealing with here)
That's the cynical view. The non-cynical view of most high end is having philosophies instead of facts and esoteric parts instead of circuitry. Oh no that's still cynical. But there's a point. If one cared to look inside most high end equipment you'd find that whatever they lacked in knowledge they sure made up in expensive content. Heck, some buy up loads of obsolete parts under the presumption that they're better than the new stuff and on the other end of the scale some have their own power transistors made in an esoteric new technology. These guys are spending money, not raking it in.
Leaving aside the question of how well-designed the innards of an audio products are, there are some very basic economic facts:
1) Regardless of market, the mark-up between the BOM and the end-user price is around a factor 5. Most of that goes to the retailer. Then the distributor. The remaining pittance goes to the manufacturer.
2) Economy of scale: production cost drops with the logarithm of quantity until you hit the raw materials cost (i'm told that the old Philips TV factory in Bruges could reliably estimate the cost of a TV set by weighing it). The same product, manufactured in tens or in thousands ends up with a markedly different price tag.
3) People want their money's worth. You spend 5k on a piece of kit, it has to look expensive. I'm talking fit and finish, not looks. It may look "butt ugly" so long as it's polished like a baby's bottom (with only as many visible seams).
So you want to make a "better than average" product? That will make it more expensive than average. So you'll sell fewer of them. So your manufacturing cost will go up and you need to increase your profit margin. The price starts getting a bit "exclusive". So people will not want the same folded casework as a $25 DVD player. Some of these look pretty sleek with the sort of plastic casting that's available when you make 100k units. If you want to get the same quality with 1k units, it'll have to be machined. This spiral continues until a company almost but not entirely runs out of customers and equilibrium is reached. Of course, that leaves room for many more small companies with slightly different offerings who appeal to a different but equally small subset of potential audio buyers. All it takes is the above 3 obvious truths to explain why the high end market is saturated with innumerous tiny companies trying to be slightly different from one another. The same spiral explains why the middle segment (where say 1500 euros would buy a very decent stereo) has pretty much vanished from the market.
People seem to have the impression that manufacturers of high end gear are greedy bastards who are making way too much money. I can assure you that the companies that churn out cheapo DVD players have a boardroom full of way richer guys, none of whom actually give a damn for audio. Even a middle manager may fetch rather more than some of the people manufacturing the high-end gear that's meeting with such opprobium for their price tag.
You can have me ranting about the complete lack of technical sense shown by equipment designers and whatnot, but as far as economics goes, the smart ones and the stupid ones are in the same boat.
trying to be different from each other. sure, an extremely fast buffer, directly connected to the output of a WM8741 is as different as it gets.
(this is about high-end pricing but touches some some aspects we're dealing with here)
That's the cynical view. The non-cynical view of most high end is having philosophies instead of facts and esoteric parts instead of circuitry. Oh no that's still cynical. But there's a point. If one cared to look inside most high end equipment you'd find that whatever they lacked in knowledge they sure made up in expensive content. Heck, some buy up loads of obsolete parts under the presumption that they're better than the new stuff and on the other end of the scale some have their own power transistors made in an esoteric new technology. These guys are spending money, not raking it in.
Leaving aside the question of how well-designed the innards of an audio products are, there are some very basic economic facts:
1) Regardless of market, the mark-up between the BOM and the end-user price is around a factor 5. Most of that goes to the retailer. Then the distributor. The remaining pittance goes to the manufacturer.
2) Economy of scale: production cost drops with the logarithm of quantity until you hit the raw materials cost (i'm told that the old Philips TV factory in Bruges could reliably estimate the cost of a TV set by weighing it). The same product, manufactured in tens or in thousands ends up with a markedly different price tag.
3) People want their money's worth. You spend 5k on a piece of kit, it has to look expensive. I'm talking fit and finish, not looks. It may look "butt ugly" so long as it's polished like a baby's bottom (with only as many visible seams).
So you want to make a "better than average" product? That will make it more expensive than average. So you'll sell fewer of them. So your manufacturing cost will go up and you need to increase your profit margin. The price starts getting a bit "exclusive". So people will not want the same folded casework as a $25 DVD player. Some of these look pretty sleek with the sort of plastic casting that's available when you make 100k units. If you want to get the same quality with 1k units, it'll have to be machined. This spiral continues until a company almost but not entirely runs out of customers and equilibrium is reached. Of course, that leaves room for many more small companies with slightly different offerings who appeal to a different but equally small subset of potential audio buyers. All it takes is the above 3 obvious truths to explain why the high end market is saturated with innumerous tiny companies trying to be slightly different from one another. The same spiral explains why the middle segment (where say 1500 euros would buy a very decent stereo) has pretty much vanished from the market.
People seem to have the impression that manufacturers of high end gear are greedy bastards who are making way too much money. I can assure you that the companies that churn out cheapo DVD players have a boardroom full of way richer guys, none of whom actually give a damn for audio. Even a middle manager may fetch rather more than some of the people manufacturing the high-end gear that's meeting with such opprobium for their price tag.
You can have me ranting about the complete lack of technical sense shown by equipment designers and whatnot, but as far as economics goes, the smart ones and the stupid ones are in the same boat.
trying to be different from each other. sure, an extremely fast buffer, directly connected to the output of a WM8741 is as different as it gets.
Last edited:
Overhere, it is common to capitalize someone else's name, not your own.
(called arrogant without an attitude)
The most disrespectful word in the English language is 'you'.
Much more pleasurable to read, Mr. P-P.
- Status
- Not open for further replies.