A good test of small condensers

rob aylestone said:
Oxygen free cables? That one i rule out before i start. I like the idea of putting the mic into a dead case though. What worries me is that measuring the noise accurately would have to be against an unknown level. If we see on the meters, say, -70dB, i have a feeling that this is uncalibrated. I am never actually convinced 0dB, the maximum possible digital level really is 0dB, because accidental overs often when examined dont actually flat top. The meters go red, but perhaps indicate red for danger when close, to warn. I’ve never tested or proved this, but i suspect it to be the case.
Click to expand...
I agree Rob, O2 free cables are snake oil. If you look up the ACTUAL science of Copper you will find that a a percentage of Oxygen in Cu is necessary for best conductivity.

I have to get ready for some heart related tests at the docs today but when I return I shall give some thought to the MO of noise tests!

An XLR male plug with a 200 Ohm termination will be required. A short across pins 2 and 3 will do but is not optimum.

Dave.
 
Julian Krause does noise testing on his YouTube channel and he has a video about how self-noise of condenser mics basically eliminates any concern about preamp noise :). Here’s a couple old links but there might be newer info on his channel. He has produced A LOT of reviews



 
I think I might well be steering clear of this one.

I've just got a 'Sennheiser' MK4 mic from China - I paid £36 for it including shipping. I have one in the store somewhere. I'll pull it out and do a comparison. Not plugged it in yet, but the box, packaging and accessories look identical. I'm seriously starting to think that big name manufacture in China is just not secure, and many of the 'counterfeits' might actually be genuine but going out of the back door? There's too much tooling and cleverness involved for it to be economic to sell so cheap?

It's another to add to the don't buy from ebay or Amazon list.
 
Those Krause YTs are interesting but I am not sure we can say ALL capacitor mics are noisier than ALL AI pre amps?

My own noise tests were rather more simple. I would use the XLR termination as he shows (but 200R since that is EU standard) and then setup a mic at say 300mm and adjust gain to get -20dBFS ish* with some "Peter Piper" record 10S waffle then be as quiet as possible for 10. Leave gain setting and swap mic for XLR term' Record 10S.

Plug mic back in and take into bedroom (Iive in a bungalow) which is very quiet anyway then wrap the mic in a duvet. Record another 10.

That process tells you the basic noise level of the pre amp when set for a high but fairly 'normal' gain. Also a reasonable idea of the mic's self noise together with the noise when you are quiet. i.e. ambient noise. For most of us THAT is the main noise source.

*Does not need to be a precise level, a few dgrs one way or tother will not alter the pre amp noise level.

Dave.
 
I stuck a label on the MK4 from China, but I cannot detect any difference in sound at all from it to the 2-3yr old MK4 I bought second hand. The box is different, the draw string pouch is a shure sized stiff zip up, and there was no clamp. Mic wise they seem identical. 4 grams difference so insignificant. My conclusions are that the German manufactured MK4 and the Chinese MK4 are too close for comfort. Or, of course, the one I bought before is also a copy. I would be quite happy using them. There is one tiny difference. on a 1-10 scale on the gain, the fake needed 1 extra point of gain to be the same? Significant? I don't know.
 
I'm surprised that Sennheiser might be making MK4s in China. Their website specifically says they are made in Germany. I don't know how the laws are about declaring country of origin in the EU, but I guess they could be subcontracting parts from China and then doing a final assembly in Germany.

The MK 4 is made in Germany, its transducer being manufactured in the same clean room as all high-end Sennheiser condenser capsules.

I have read some comments about the MK4 sounding quite like one of the lower end Neumann mics.
 
I've tied myself in knots trying to get some sensible noise figures. It got very complicated - BUT - I have some data, but it is very confusing.

Please point out any holes in how I have done it, or wrong methods.
Mic - I chose a Coles lip mic so identical distance from my mouth every time. Not the best tone, of course, but consistent.
Interfaces available. I got a Tascam 1641, a Presonus FP10 Firepod, a very cheap Chinese red Focusrite copy, a Tascam H6 and the video camera preamp, which embeds audio into the SDI file and records onto a Blackmagic Hyperdeck.
In-line preamps - two laying around here, one very cheap (less than £10 cost) and a better one (just under $50) - neither anything special.

Process - we all know that flat out on interface/pre-amp is worst for noise, so I used the sweet spot on what I have available - typically on 1-10, 7.5 or 8 in practice fr low output mics (and the lip mic is very low output.

This means that the actual level recorded is below optimum - sometimes quite a bit less. I then normalised each file to 0dB, and in the tools in Cubase I could read the levels. What I did was speak, then record silence - well, taking the mic away from my lips in a pretty quiet studio. This gave me the noise floor figure. It seemed to actually vary a little, so I did three spot readings in the silence and averaged them.

In some cases, I added or removed the in line preamps and got a set of results that were repeatable. None of the averaging of the three results was out by more than 2dB.

As far as I can tell from the commentary of what I recorded, the results are accurate - within the test method.

As an aside, it meant connecting the firepod to something with firewire - which was an imac running catalina. Guess what? Not supported. I ended up finding a third party driver a bit like asio4all, from apponic.com - which actually worked!

alt=


The best performance was from the Presonus - which had the noise at -54dB. This seems quite poor as a figure. the lowered gain on it, plus the in-line preamp was the result I was expecting, I just hoped for a bit better performance. without the in-line pre, and turning the gain up produced -52dB - which is what we have often considered normal on pre-amp noise topics. The Chinese interface was best gain full, no in-line - this makes no sense? Adding full gain put it 4th? Turning the gain DOWN increased the noise. I double checked and adding normalisation definitely produced a noisier result. I'm not surprised by the camera audio - pre-amp on camera, then conversion to the embedded audio. This then gets stripped out of the hyperdeck video file, which is a bit of a conversion path, and the camera preamps are not designed for low level inputs really.

It's taken ages to do this - but I'm not sure what, if anything, the data proves? There must be mismatches somewhere - so maybe Dave's impedance matching/loading? I don't know. I'm slightly dubious of the actual measurements as the noise visually went up and down rapidly in the 'silence' and I used some of the adjustments to slow down rate of change influencing the readings. This may or may not have influenced the result. The readings also provided a peak hold which during the silence could take the averaged -50dB figure and have occasional peaks of -30dB. I don't know what is in the waveform to give these peaks - I hear silence? (well, the hiss).

what do you think? Does it mean anything?
 
rob aylestone said:
I've tied myself in knots trying to get some sensible noise figures. It got very complicated - BUT - I have some data, but it is very confusing.

Please point out any holes in how I have done it, or wrong methods.
Mic - I chose a Coles lip mic so identical distance from my mouth every time. Not the best tone, of course, but consistent.
Interfaces available. I got a Tascam 1641, a Presonus FP10 Firepod, a very cheap Chinese red Focusrite copy, a Tascam H6 and the video camera preamp, which embeds audio into the SDI file and records onto a Blackmagic Hyperdeck.
In-line preamps - two laying around here, one very cheap (less than £10 cost) and a better one (just under $50) - neither anything special.

Process - we all know that flat out on interface/pre-amp is worst for noise, so I used the sweet spot on what I have available - typically on 1-10, 7.5 or 8 in practice fr low output mics (and the lip mic is very low output.

This means that the actual level recorded is below optimum - sometimes quite a bit less. I then normalised each file to 0dB, and in the tools in Cubase I could read the levels. What I did was speak, then record silence - well, taking the mic away from my lips in a pretty quiet studio. This gave me the noise floor figure. It seemed to actually vary a little, so I did three spot readings in the silence and averaged them.

In some cases, I added or removed the in line preamps and got a set of results that were repeatable. None of the averaging of the three results was out by more than 2dB.

As far as I can tell from the commentary of what I recorded, the results are accurate - within the test method.

As an aside, it meant connecting the firepod to something with firewire - which was an imac running catalina. Guess what? Not supported. I ended up finding a third party driver a bit like asio4all, from apponic.com - which actually worked!

View attachment 143885

The best performance was from the Presonus - which had the noise at -54dB. This seems quite poor as a figure. the lowered gain on it, plus the in-line preamp was the result I was expecting, I just hoped for a bit better performance. without the in-line pre, and turning the gain up produced -52dB - which is what we have often considered normal on pre-amp noise topics. The Chinese interface was best gain full, no in-line - this makes no sense? Adding full gain put it 4th? Turning the gain DOWN increased the noise. I double checked and adding normalisation definitely produced a noisier result. I'm not surprised by the camera audio - pre-amp on camera, then conversion to the embedded audio. This then gets stripped out of the hyperdeck video file, which is a bit of a conversion path, and the camera preamps are not designed for low level inputs really.

It's taken ages to do this - but I'm not sure what, if anything, the data proves? There must be mismatches somewhere - so maybe Dave's impedance matching/loading? I don't know. I'm slightly dubious of the actual measurements as the noise visually went up and down rapidly in the 'silence' and I used some of the adjustments to slow down rate of change influencing the readings. This may or may not have influenced the result. The readings also provided a peak hold which during the silence could take the averaged -50dB figure and have occasional peaks of -30dB. I don't know what is in the waveform to give these peaks - I hear silence? (well, the hiss).

what do you think? Does it mean anything?
Click to expand...
I don't understand the correlation between these interfaces and SDC’s?
 
None at all, just a progression as the suggestions rolled in. What intrigued me was the need to consider what the preamp sections could introduce. A valid point I’d sort of dismissed. What i expected was confirmation of basic things in the noise figures, and it didn’t quite work as I expected.
 
rob aylestone said:
None at all, just a progression as the suggestions rolled in. What intrigued me was the need to consider what the preamp sections could introduce. A valid point I’d sort of dismissed. What i expected was confirmation of basic things in the noise figures, and it didn’t quite work as I expected.
Click to expand...
I might be able to shed maybe a glimmer of light on your findings Rob?

The 'odd' good result for the Chinese mic sans pre is I think due to it having an exceptionally low noise circuit (does not mean it has good headroom!) The way most pre circuits work is that there is a pot of between 5k and 50k that adjusts the gain. At max gain the value is very low, maybe just 10R and so its Johnson noise is tiny. As you reduce gain, more 'noisy' resistance is inserted.

The fact that the bare noise figures look poor is I think a result of the normalization to 0dBFS? That has not changed the relative noise levels but has made them look bad. What is needed I feel is the signal to noise RATIO between sound source and the various noise mechanisms? Tricky!

Dave.
 
Thanks Dave. Bring things up to 0 was the only way I could think of to measure down, if you see what I mean. Clearly, it's just very difficult to produce results that make sense. It was interesting. I'm thinking that perhaps the only way to get meaningful comparisons is to NOT change the studio layout or the sound sources. People sometimes get annoyed with me for using the SM7B at a distance, and the doing the same thing with every mic. One today said that X mic needed to be used close in to judge it. Valid point, but the way they all react differently to proximity effect colours the results. Using them at a similar distance allows the 'tone' of the mic to be compared with another. I've done a video with 414s and some others on my piano - so maybe I just need to revisit how I did that and replicate it with new mics - then the old and new video make sense listened to, together?

If I do it differently, then the comparison will be flawed?

So far, the mics have not appeared and I can't really actually ask for them. There is also an HD24 somewhere, but I don't feel comfy asking for it, and I don't need it.
 
Rob, if you still have the original recordings you could export them as 16 bit* .wav and run in the RTA of Right Mark Analyser. That will give you a series of plots and if you generate 1kHz at say -20dBFS you can see the levels relative to that. You will also see any 'nasties' such as hum or interference.

*My version of RMA won't run 24 bit files but a later version might now?

Dave.
 
My 184 test is scrapped! The mics arrived and are in a nice wooden box and look nearly new. Plugging them in reveals a mess. I think they have been stored in a damp garage for maybe two or three years. I tried the usual warm up and bags of sillica gel, but the noise floor is really high and after cleaning all the mating surfaces I've given up and returned them as a 'momento' to keep rather than mics that can be used. Far too noisy. One is worse than the other. The internals are dull and almost look like they'd been in a flood? The outsides look perfectly normal but I can't do anything with them. On one side of the board are two capacitors (other side to the multipin IC) They look really dull and like they have actually been wet at some point? Oh well. never mind.
 
rob aylestone said:
My 184 test is scrapped! The mics arrived and are in a nice wooden box and look nearly new. Plugging them in reveals a mess. I think they have been stored in a damp garage for maybe two or three years. I tried the usual warm up and bags of sillica gel, but the noise floor is really high and after cleaning all the mating surfaces I've given up and returned them as a 'momento' to keep rather than mics that can be used. Far too noisy. One is worse than the other. The internals are dull and almost look like they'd been in a flood? The outsides look perfectly normal but I can't do anything with them. On one side of the board are two capacitors (other side to the multipin IC) They look really dull and like they have actually been wet at some point? Oh well. never mind.
Click to expand...
Probably not worth investing in, but might you consider a rebuild/restoration?
 
The trouble is its a tiny board but with loads of surface mount components and thats now beyond my eyesight, dexterity and equipment level. She has a nice box containing his old bits. I dont think ai really need them.
 
Oddly, i swapped the capsule between the noisy and very noisy one, and that didnt change things, so I suspect the problem is on the PCB?
 
rob aylestone said:
Oddly, i swapped the capsule between the noisy and very noisy one, and that didnt change things, so I suspect the problem is on the PCB?
Click to expand...
***t or bust try for contaminated PCBs.
Wash with deionized water. Hang in airing cupborad for a week. Soak in WD-40, let drip off a day. Give it a do.

NO promises!

Dave.
 
You must log in or register to reply here.
Back
Top