How do I measure head response correctly?

[budharpey]

Hello,

I studied quite a bit in this great forum now and found the very useful thread of markrob about his DIY Recording Head: https://www.lathetrolls.com/viewtopic.php?f=15&t=3027.
I have some questions referring to the issues mentioned in this thread, but as it already is quite long and discusses various matters I thought I'd rather start a new thread for my kind of "basic" questions.

1) first problem: how to measure the head frequency response correctly?

My idea would have been, to record some piece of audio (either a generated frequency signal with range from maybe 20 hz up to 12000 hz or simply a music tune) with the recording head on polycarbonate and than play it back with input to True RTA with the "Peak hold" option. Then I would see the frequencies which were recorded. One would have the Anti RIAA filter in action while playing the record back, and this way the spectrum should be more or less a horizontal line until the upper frequency border, which would be depending on head quality somewhere between 8 and 12 kilo Hz (disregarding the resonance peak somewhere in the spectrum). Problem with this way of analyzing would be that undesired noise which was recorded and played back would be also in the analyzing spectrum.

In the above mentioned thread markrob writes that he used a "Kaman Eddy current position sensor" and obviously TrueRTA as software to measure.
I wonder how this works, I can't find out which kind of sensor model this should be and how to apply it to the recording head and what source signal to use.
How is the frequency plot in the post a displacement vs. frequency plot? As I understand the plot is a frequency response of some source signal sent to the head, right? But if it isn't played back with anti RIAA filter, why is the line so horizontal if the source signal was equalized?

2) second problem: with this kind of DIY-Stereo-Heads in combination with steel embossing styli, how would I want to record to achieve loudness?

a) Would I use as much weight as possible or as little as possible, or something in between (eg how many grams)? I can see a huge weight placed over on the head in the last picture of the first post in this thread?
Before, I thought I would use as little weight as possible to have the maximum of needle movement = amplitude = loudness?

b) Will I want to record really loud (just before the speaker will begin to distort)? I don't really understand the correlation of speaker wattage and amp wattage. In the used speakers in the thread are 5 watt speakers, but markrob uses 60 watts per speaker to record - is this due to equalizing anti RIAA, as to speek the low frequencies are cut out and the speaker is capable to take more watts? How do I know how much watts a speaker can take in correlation with equalizing?

3) third problem: How is the mentioned ferrofluid applied and used to damp the heads resonance?
I don't know anything about ferrofluids so far, as I understand it, it is a fluid which reacts to magnetism. How can one apply this to the speakers in the head? What kind of product would one buy?

Sorry for this kind of questions which might seem "rather dumb" to some of you.. I am a young computer science student from Germany who likes to tinker. I have learned basic knowledge of feedback control systems and electrical engineering. Sadly i don't really have consolidated knowledge in acoustic engineering so far. I have made first experience recording on polycarbonate with a rude first mono lathe attempt (https://www.youtube.com/playlist?list=PLVF0zYzd7tVQPDo-wu8OmxqdwMjkz-YAd) and now I would like to go on to build something more useable for home use... It would be awesome if some of you could help and clear things up a little bit..

budharpey

[dubcutter89]

Grüß dich!

Zu deinen Fragen:

1) head frequency response: there are a lot of different methods...
as you already said you can record a frequency sweep, play it back and then look at the amplitudes
if you use inverse RIAA for recording and normal RIAA for playback the curve ideally would flat. normally the noise floor is way below the music signal, so maybe first try to improve signal to noise ratio:P

you could also use an old pickup and glue it to the tip of your cutter, the rest is the same as above...

the elegant way is to use a method called light-beam-pattern
here you would record a freq. sweep with no equalization (ie. constant voltage across the head)
when you look at the record with a torch you should ideally see reflections of equal thickness (or if you use recording eq the reflections look like a christmas tree)

you could also use a scope and measure the excursion of the grooves, but that is only good if you have too much time...
or maybe you have a laser interferometer...

2) try and error
more watts equal more volume - the upper limit is when your head burns out! (or your playback system skips)
standard cutters (not moving iron) can usually handle 1-5W continous, but you want some headroom for eq and transients and therefore should use something more powerfull. 50-500 Watts are common...

3) Either buy ferrofluid speakers (look for tweeters), or you could get some and try to get it into the magnetic gap

hope this helps

Viel Erfolg!

Lukas

[markrob]

Hi,

One approach I've used is to record some program source material with the RIAA in place. Capture the playback of the cut via a turntable (with RIAA) to a WAV file. Use a program like HAR-BAL or Voxengo Curve EQ

http://www.voxengo.com/product/curveeq/

To match the source (without RIAA) of your cut to the playback. The EQ curve you develop is then applied to your source material and the processed file is again cut (with RIAA). Within the physical limits of the head, you will now have a good match. Once you obtain the EQ curve, you can use it on other source material and you should be in the ballpark. If you use this approach, don't ttry to push the highs too far or you will find that you will need too much power or have to reduce you cut level too much to avoid smoke (see below).

If you use this method, you will see that you will need quite a bit of correction to flatten the head response (as much as 30-40db). Due to the large amount of high frequency boost coupled with the RIAA, you will need a good deal of reserve power to cut the high frequencies. Use as large a power amp as you can find (100 watts as a minimum). However, you MUST still keep the long term average power within the heads' power handling capability (5 watts or so). You can allow short 100-400 watt peaks, but the average needs to be monitored carefully or you will have smoke. With that in mind, cut at low levels at first to see if you are in the right ballpark as far as fidelity goes. Then ramp up the power. Don't expect to cut at commercial levels with a DIY head.

One other point worth mentioning is that the open loop mechanical stiffness of the head at the cutting stylus is such that it is influenced by the cutting medium. So you'll find that the head response on the bench will differ once you cut. It will also change if you cut poly or lacquer. Since this can and will vary over time, your EQ will probably need to be tweaked a bit for each cut. That's the big advantage of feedback as it will compensate for these changes automatically.

You could create a head design that is much stiffer than the media (which would reduce or remove this dependence), but the efficiency of the head would really suffer (much more drive needed).

Hope that makes sense.

Mark

[budharpey]

Vielen lieben Dank Lukas & Thanks Mark for taking the time for your quick and detailed response and sharing your knowledge!
It explained already a lot to me! I am excited to try and test some of the methods and software mentioned over the next week and will be back then to report or maybe to ask, if further questions appear in the process...