Fair play. Nothing wrong with the other S.. ranges though... 30V SC. Datasheet requires 1uf 16V. Plenty of options except polyester.
I lost the point of this discussion. Are we talking about the gain setting caps normally relegated to a simple 1uF x7r SMT? Why is it critical to have a good low ESR cap here? The gain is only set one time at power on - the internal ADC reads the voltage set by the two resistor voltage divider and sets the internal gain to one of 5 values and there it stays until next power on cycle. Or am I totally confused as to what you guys are discussing?
Oh I love tantalum's......nothing like the unsurity of whether the amp will explode every time its powered up!!!!
Crossed fingers. Have yet to experience a Chinese board melt down - except in the aforementioned Dell SX280 computers that I had. It's an interesting story actually as 6 of 8 of my parallel cluster had exploded power caps exactly 1 month past warranty expiration. I was kind of upset so did some research. It turns out that the caps were all made by a company in China and Taiwan who hired one of the defecting designers of the latest high performance electrolytic caps from a leading Japanese company. He took with him the recipe for the latest caps but apparently took a flawed recipe (perhaps his employers were onto him regarding the imminent defection?). Anyhow the new company spun up a whole production line and made and sold products that blew up after so many hours. There was a class action lawsuit and this all came to light. I was still stuck with 6 dead motherboards oozing with cap goop. I think this cost Dell $0.3B in the end for buying caps made with a bad recipe.
http://www.theguardian.com/technology/blog/2010/jun/29/dell-problems-capacitors
http://www.theguardian.com/technology/blog/2010/jun/29/dell-problems-capacitors
Last edited:
gvdd is used by resistor divider to supply gain setting to plimit/gain slave pins. But I thought gvdd was the supply for the supply for the gate drives? And isn't the cap, tied in with the voltage divider, the gvdd to GND? Datasheets suggest class2 ceramic for that.
I would presume a GND cap for gvdd, tied to the ldo along with avcc would require quality replacement as much as bootstrap tied to gates at other side! Maybe I am reading the wrong data sheet!😕
Since it is class D, gate drive voltage just needs to be digital stable for on/off operation. Not sure if higher quality caps will affect performance of the gate from switching properly. The AVcc on the other hand is a true analog voltage used by the analog comparator for the PWM. It is critical and hence the suggestion by some to de-couple pin 17 from main power rails and give it its own bypass caps.
Come to think of it, maybe this is why the TDA749x requires an external linear voltage regulator IC for 3.5v? Maybe it is used as a clean source of analog voltage for the comparator? Maybe that would make it have inherently lower noise performance?
Come to think of it, maybe this is why the TDA749x requires an external linear voltage regulator IC for 3.5v? Maybe it is used as a clean source of analog voltage for the comparator? Maybe that would make it have inherently lower noise performance?
The gvdd provides the current to set gain through the resistor divider. Caps allow AC through to GND. Better grounding caps!
From the data sheet schematic of the IC internals it looks like gvdd goes internally from LDO to the gate drives, and the bootstrap lines joint it after diode. Its this line the external gvdd out pin connects internally, and shoots to GND.
Guess I need to learn more about LDO's, Gate drives, high side switching, voltage, current, GND and loops!🙂
From the data sheet schematic of the IC internals it looks like gvdd goes internally from LDO to the gate drives, and the bootstrap lines joint it after diode. Its this line the external gvdd out pin connects internally, and shoots to GND.
Guess I need to learn more about LDO's, Gate drives, high side switching, voltage, current, GND and loops!🙂
Exactly. The 3116 avcc flows through LDO, and it outputs LDO alongside gvdd. The gvdd is taken out of the LDO which is powered through avcc.
The external LDO versions keep the gvdd supply separate to the avcc. Simples....as said by the merecats😎
A MKT capacitor has:
- Extremely low leakage. Insulation strength is tens to hundreds of gigaohms for a 1uF part. It'll hold charge for months.
- Low ESR. Not quite as good as MKP and some other film types, but still respectable.
- Low ESL, largely determined by physical size/pin spacing.
- Some temperature variation of capacitance, but provided you're not using it in an application where that matters (eg, an audio filter, timer circuit, etc) then you're fine.
Open up a high power switching power supply like something used for telecom or an on-line UPS, you'll find these caps on the front of the main converter. Reason being they can handle many amps of ripple current with 400V volts across them - you'll only blow up electrolytics in this application, and ceramics don't scale to sufficient uF's to be usable. They make great decoupling capacitors.
That being said, for GVDD decoupling I'd go with a ceramic, I use a 1uF/25V X7R in an 0603 package on the Wiener card. Reason being the ceramic has low ESL/ESR, and can be placed very close to the chip limiting additional "layout-induced" ESL/ESR - providing a low impedance at the GVDD pin well up into the tens of MHz.
That's the cap I am talking about.
Thank you. I just asked if anyone had tried other than x7r! Class1 don't scale, but many rate mica or tantalum above x7r.
I did see in original discussion on here regarding bootstrap mod a suggestion to try tantalum's there, so wondered if anyone tried here.
Don't get me wrong, I like polyester caps, not as much as polystyrene and teflon, but more than PP. Then again I only use them in signal lines.
Thank you. I just asked if anyone had tried other than x7r! Class1 don't scale, but many rate mica or tantalum above x7r.
I did see in original discussion on here regarding bootstrap mod a suggestion to try tantalum's there, so wondered if anyone tried here.
Don't get me wrong, I like polyester caps, not as much as polystyrene and teflon, but more than PP. Then again I only use them in signal lines.
Last edited:
MKP, MKT, ... ???
For those unfamiliar with the naming convention on film caps, here is what I found that was very helpful. It has to do with the Germans being the inventors of the metalized foil film cap...
From: MKT, MKP, MKC, KT, KP, KC,... - 8052.com:
For those unfamiliar with the naming convention on film caps, here is what I found that was very helpful. It has to do with the Germans being the inventors of the metalized foil film cap...
From: MKT, MKP, MKC, KT, KP, KC,... - 8052.com:
Tantalum != Tantalum..
Beside this:
https://en.wikipedia.org/wiki/Conflict_resource
I'd never use tantalum on a bootstrap.. (beside that i wouldnt use tantalum at all)
As i asked in the past, what are you optimizing for? (And why noone does some measurements or math regarding a "given" problem to solve)
It's always like "I heard/someone suggested.. Wouldnt it be better... Does this cap "sound" better than..."
Cap rollin..
Makes me sad.. 🙁
What needs to be improved - THD?
Whats the actual parameters - 0.5% 0.05% 0.005%?
At what VCC - 10V, 20V, 24V?
What power supply - SMPS (primary/secondary, flyback, frequency, output resistance), Battery (SLA, Li+)...
At which freqency - 1kHz, 5kHz?
Load resistance - fullrange speaker, 2-way speaker with passive x-over?
Cabinet - closed baffle/open baffle, reflex?
At what output power - 1W, 100mW, 10W?
Crest factor?
etc........
-.-
Last edited:
First answer would be that despite huge human rights violations, everyone seems to be sources parts/boards from China still. So a tiny bit of natural material doesn't weigh heavy on my shoulders.
Secondly.... Hate look of stock caps. Don't want polyester there. When I touch any of the stock "blue box" caps the result is noise. So if I am to "redo" solder joints I'll replace with a cap I can identify make, model and data on, unlike the stock caps.
Basing the theory on every piece of documented evidence and knowledge others well respected in industry have been kind enough to publish.
As for measurements, not everything in audio shows up in measuring, not all measurement differences are audible.... So I am wary of such nerdishness. I follow thought processess, experimentation, and listening experiences. Plus I hate to be a lemming.😉
Like you do as well?
But:
...
So buying a cheap chinese board, with "cheap chinese" caps, which needs completely overhauling is "a tiny bit of natural material"..
😕
What are you doing with the discarded caps, put em into waste? So, because of the demand, these caps where directly manufactured for wasting em?
Hmm..
Heard that alot in the past. Seems like you only trust in things you can hear. For sure, your ears are fully unbiased. (not) I mean, when using like a $100 coupling cap and knowing it, wouldn't there be a prebiased expectation about the results?
But back to topic. 😉
Then the method of measuring/setup is wrong.
True, thats why someone should measure. Tuning parameters and/or part-specs which aren't audible are (imho) waste of money/energy/ressources when all other aspects (like EMV/EMI) are met and within reasonable specs.
Regards, doc
Last edited:
So buying a cheap chinese board, with "cheap chinese" caps, which needs completely overhauling is "a tiny bit of natural material"..
😕
What are you doing with the discarded caps, put em into waste? So, because of the demand, these caps where directly manufactured for wasting em?
Hmm..
Heard that alot in the past. Seems like you only trust in things you can hear. For sure, your ears are fully unbiased. (not) I mean, when using like a $100 coupling cap and knowing it, wouldn't there be a prebiased expectation about the results?
But back to topic. 😉[/QUOTE]
Dont deny I buy from China, Taiwan, and even USA. That's cos too expensive in UK to make or buy components. Its economics.
The discarded caps and resistors, other components as well from old stereos, all go into a
Parts draw for when I test things out. For example, while I was nailing down the value and model of input caps for 3116 boards I used old parts to mock up pllxo, and the 1uf caps were bypassed by nichicon 4.7uf caps from an old 1990's stacking stereo system I broke up a few months back thrown out in neighbours skip.
As mentioned, I buy from abroad to avoid stupidly expensive " audiophile" caps etc. I'm not putting oscons in as DC coupling caps, or black gates, or nichicons etc. I've relied on caps I've been pleased with before and cost me $1.89 for 12 caps shipped from China.
I've taken part in system builds for near 20years. Research, experiment, finalise, build. If not happy then strip back and rebuild! At each stage I listen, not measure. I listen, and compare what I hear with a range of other devices I've heard playing same thing.
Of course I use common sense as well. Example?.....why put expensive cap with low ESR in series with coil, or even as 3116 input cap with 30k resistance.
Its a trade off between electrical engineering, common sense, and real audible results.
I do listen to what I hear, or read, but from listening I question, cross reference, and make informed choice.
But ho hum.
Doctormord.... I am not arguing back. I know all too well I write abrasively, and, far to often for some, as honestly as my mind thinks.
As for the gvdd GND cap, I don't personally think a polyester cap is best used for bypassing to GND. I realise the gain is set at power up, but LDO is powered by avcc, and it is out of LDO that gvdd comes out to gate drives, and passes to GND.
I've researched various op amps, gate drive, LDO, other tpa's and ti docs, as well as research by manufacturers such as avx/kemet etc on ceramic vs tantalum and mica, as well as Elliot sounds & eetimes etc etc etc. Many point to the cap needing to be as clean, efficient, and low ESL as you would expect any other signal or power to GND. The duty of the cap is out there in the documents, and the relevant requirements also. Its just piecing the clues together.
Anyhow. My apologies for abrasiveness.
As for the gvdd GND cap, I don't personally think a polyester cap is best used for bypassing to GND. I realise the gain is set at power up, but LDO is powered by avcc, and it is out of LDO that gvdd comes out to gate drives, and passes to GND.
I've researched various op amps, gate drive, LDO, other tpa's and ti docs, as well as research by manufacturers such as avx/kemet etc on ceramic vs tantalum and mica, as well as Elliot sounds & eetimes etc etc etc. Many point to the cap needing to be as clean, efficient, and low ESL as you would expect any other signal or power to GND. The duty of the cap is out there in the documents, and the relevant requirements also. Its just piecing the clues together.
Anyhow. My apologies for abrasiveness.