Yes, you do (IMO) - with transfomer input theperformance of the amp 'improves' quite a lot - everything sounds more transparent/open/natural - once done/experienced, I cannot imagine myself (or anybody else for that matter) going back.
In theory, this is correct.
However, I cannot stress enough that you need to make damned sure that the source device outputs no DC before doing any "testing".
I made the mistake of reading some info about a device (a popular, affordable DSP) in another forum and decided to go ahead and remove the input caps on one of my class D amps.
There was a slight "hissing" sound and thankfully I was close enough to my right speaker (open baffle) that I caught that nasty smell in my nostrils within a second or 2, and then pulled the plug immediately. 😱
Thankfully, the drivers (MarkAudio Alpair 10P) still sound excellent. I am sure I have shortened their lives, though. 😡
Not a fun way to be reminded of two very important lessons in DIY audio; 1) do not believe everything you read on the internet and, 2) use cheap drivers for testing (or even just a multimeter!)
Last edited:
You are talking about using a transformer instead of input caps, but ChrisMmm is asking about removing the input caps (replacing with straight wire, not a transformer). That is what I am warning about.
The transformer removes DC. We get that.
Sharpi,
Thanks for the tip! These look ideal and the price is indeed in the $10 to$15 ea range. Here is a nice PCB stereo version with lots of impedance matching options from same company for $31.
https://www.edcorusa.com/umm-9pc
An externally hosted image should be here but it was not working when we last tested it.
Input Maximum Voltage (20Hz.) 10Vrms
Input Maximum Voltage (40Hz.) 20Vrms
Input Impedance 10K, 15K, or 30K Ohms
Output Impedance 10K, 15K, or 30K Ohms
Frequency Response 20~20K Hz., <1dBu
THD+Noise <0.05% @ 1K Hz.
Bobbin Material Glass Filled Nylon 6/6
Flamability Rating Class B 130°C
Core Material M6 29 ga. line grain oriented steel
Termination Eurostyle connectors
Mounting Surface Mounted
Weight 0.4 lbs.
Compliance RoHS & REACH
Well, some "audiophiles" insist that transformer coupled amps sound a lot better than capacitor coupled amps. They argue that any capacitors in the signal path, especially at the input will degrade the sound. A few very expensive tube preamps are transformer coupled and the transformer is wound with silver wire.
Many arguments that have been put forth for transformer volume control - better isolation, minimal impact on the dynamic range at low signal level, convenient impedance matching, are applicable for transformer coupled input. Actually, an easy experiment would be to place a TVC (transformer volume control) in front of the amps input. Hey, may be I can do that experiment when I get the YJ blue/black board for modding.
As we know, DC cannot get pass a transformer. So one is safe even though no capacitor is in place. However, DC will saturate the core quickly (as pointed out by someone earlier) and transformer will not function properly.
Regards,
I've been using ferrite (PQ20) cores and winding my own. Very cheap indeed in parts cost. Mouser has the PQ20 cores in stock but no bobbins as yet.
Primary inductance is important as it'll determine the low-frequency loading the trafo presents to the source. Other than that, turns ratio and winding configuration - best to use a centre-tapped secondary to drive the TPA.
@ChrisMmm - the purpose of using the trafo isn't to get zero DC, that's a side effect. The main purpose is to get better sound, I believe this happens because common mode noise is greatly reduced. The trafo almost takes the connecting cable out of the equation in terms of SQ.
Last edited:
I would generally only be using one source as input and as I build all my own DACs etc. I should be able to determine if they are DC free or otherwise. All my DACs that do have DC present already have DC blocking caps. My present DAC, an ES9023, states "Using an integrated charge pump to generate the negative supply, the ES9023 can operate from a single AVCC supply to drive a ground-referenced 2Vrms output, eliminating the need for output dc-blocking capacitors."
I have an autoformer based attenuator as a pre but unfortunately these do not block DC like a transformer.
But I certainly take your point and would test with cheapo speakers, I wouldn't want my Alpair 10.3s damaged either!
Makes sense, isolation as well as DC blocking. Be interesting to compare trafo input against straight wire as opposed to comparing with a cap.
Most/all datasheets specify temperature coëfficiënt and maximum tolerance for temperature. My point was NOT temperature LOL a 50V 3uF XR7 ceramic is only 3uF at 0V.
As example, you can find many also about through hole and some really good datasheet mention voltage coëfficiënt, most however don't:
Temperature and Voltage Variation of Ceramic Capacitors, or Why Your 4.7F Capacitor Becomes a 0.33F Capacitor - Tutorial - Maxim
I believe that an input transformer does the following:
- isolate source device from +3VDC bias on TPA input pins
- convert single ended sources to differential, enabling the TPA to do a better job of noise cancellation
- offer deeper bass without creating a turn on thump issue (the datasheet warns against using big input coupling caps, as an RC constant of >1ms increases the thump)
- act as a desirable low pass filter on the input signal, removing noise before it gets to the amp
- offers the opportunity to attenuate or increase the input signal, for improved matching with source device
- avoids ground loops
I'm no expert but there seemed to be sufficient potential advantage, so I tried the transformers and preferred what I heard. Where transformers seem to 'fit' nicely with these amps, coupling caps don't seem to 'fit' as nicely. The usual approach of increasing cap value above the standard 1uF to reduce bass roll off and going for a film cap for improved sound tends to mean a bigger cap, which I believe can increase the noise picked up by this component and added to the signal.
I'm sure you can get a great result with both caps and transformers. I had better success with transformers, and fully admit to being happy with the technical 'synergy' these components appear to bring, so haven't spent as much time playing with better cap configurations.
- isolate source device from +3VDC bias on TPA input pins
- convert single ended sources to differential, enabling the TPA to do a better job of noise cancellation
- offer deeper bass without creating a turn on thump issue (the datasheet warns against using big input coupling caps, as an RC constant of >1ms increases the thump)
- act as a desirable low pass filter on the input signal, removing noise before it gets to the amp
- offers the opportunity to attenuate or increase the input signal, for improved matching with source device
- avoids ground loops
I'm no expert but there seemed to be sufficient potential advantage, so I tried the transformers and preferred what I heard. Where transformers seem to 'fit' nicely with these amps, coupling caps don't seem to 'fit' as nicely. The usual approach of increasing cap value above the standard 1uF to reduce bass roll off and going for a film cap for improved sound tends to mean a bigger cap, which I believe can increase the noise picked up by this component and added to the signal.
I'm sure you can get a great result with both caps and transformers. I had better success with transformers, and fully admit to being happy with the technical 'synergy' these components appear to bring, so haven't spent as much time playing with better cap configurations.
Some questions:
If size of (both) capacitors on differential inputs is of concern, the size of transformer also is, noise gets cancelled when there is no signal, but mixes with signal when there is signal, IMD distortion??
Inputs already are differential, noise picked up by negative cap might to some extend already be cancelled by noise picked up by positive cap??
When connecting a transformer should signal to tpachip not be shielded and connected to primary side, because tpachip/psu itself might be main cause of noise??
If size of (both) capacitors on differential inputs is of concern, the size of transformer also is, noise gets cancelled when there is no signal, but mixes with signal when there is signal, IMD distortion??
Inputs already are differential, noise picked up by negative cap might to some extend already be cancelled by noise picked up by positive cap??
When connecting a transformer should signal to tpachip not be shielded and connected to primary side, because tpachip/psu itself might be main cause of noise??
What kind of audio transformer would you recommend?
If possible available from digikey.
Thanks
I concur with Sharpi's findings; for me there is no way back to whatever capacitors at the input. Quieter, more dynamic, more detailed and deeper and more focussed bass are obvious. But there are some things going on which are difficult to describe, maybe more fluid, coherent, better PRAT.
Sorry for all this audiophool vocabulary 😉
I am using Lundahl LL1540 btw, unfortunately these don't come cheap, but sometimes are offered in the bay for around 40-50$ a piece. That seems to be a lot considering the price of the board being half of one transformer, but to me it is clearly worth the stretch.
Also because this little amp sounds damned good and is getting better with every little tweak; it is frightening. Definitely a keeper with right speakers, and even expensive tube amps will have trouble to catch up with it.
I am about to build an OTL the next months, parts are sourced already. Will be interesting to compare with my ow setup here....
Sorry for all this audiophool vocabulary 😉
I am using Lundahl LL1540 btw, unfortunately these don't come cheap, but sometimes are offered in the bay for around 40-50$ a piece. That seems to be a lot considering the price of the board being half of one transformer, but to me it is clearly worth the stretch.
Also because this little amp sounds damned good and is getting better with every little tweak; it is frightening. Definitely a keeper with right speakers, and even expensive tube amps will have trouble to catch up with it.
I am about to build an OTL the next months, parts are sourced already. Will be interesting to compare with my ow setup here....
Danzz,
I tried finding something on Digikey and was unable to sift through the 15,000 offerings. I did find this unit from a company based on Sharpi's recommendation. It has 10k, 15k, 30k ohm taps and is stereo with great freq response and low THD figures. I did not look for it on Digikey but the company sells direct for $31 ea. I asked about it in post #4080 but no one responsed whether or not the specs seem good.
https://www.edcorusa.com/umm-9pc
Someone please check specs and let me know if appropriate for input transformer on 3116.
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
xrk,
These look nice. Everything is installed on a PCB and ready to go. Or you can choose from the XSM series and pick individual transformers. It cost less and gives you more freedom in installation. I have been contemplating getting a few of these to made some single end to balance input adaptors. If I remember correctly, these transformers are made to order (there are no stock).
You are right, the specs of these transformers look pretty good. Let us know how it works.
Regards,
I had a quick look and its unclear if these can be used to give a centre-tapped output (unbal in, bal out). The text at the top suggests that balanced out is possible but it doesn't say if that applies to all combinations - none of the individual line items mention balanced out. Having only impedances (10k/15k/30k) as the spec rather than turns ratios kind of leaves things a bit fuzzy.
Of course they're still usable even with only unbalanced out, but not really ideal.