Ferrofluid ! In the gap, your opinion ! And helium

[Lee2023]

Hi !

Happy holidays !

So decided not to put ferrofluid in the gap of my custom made magnetics, done some measurements before, really improving heat transfer to the body of my assembly, also reducing amplitude of my main resonance.. but still really not sure about this, how it can affect the transient speed of thing, if it will dry or change properties in time, put a lot of time in this assembly so I really cautious with decisions

Ferrofluid exist for a bit of time mid 70's supposed to, so why they doesn't used it in cutter head ? Must be a good reason, or they used it and I don't know they did... but if it was helium cooled in some heads..

Must have the same effect in tweeters and audiophile related opinions

Also for my next revision I want to explore helium cooling, one of my idea was to put some TEC with small cold plate on the heat side to bias the core of my magnetic assembly just above the dew point to gain some extra margin before my glue reach it's max working temp, not impossible but a bit crazy, but with helium how do you manage the flow not to be turbulent and ''drive'' the voice coil assembly with noise that would end up cutted on the disc ? Or the flow is just so low that it could not modulate ?


Thanks !!

Lee

[Lee2023]

Ok so for the flow rate at first was not sure that this could be find via google or AI overviews or or.. so Neumann head 0.05 SCFH around 0.0008 CFM so this explain the fact that you keep the cavity filled to allow better transfer to the rest of the assembly

[zdenek]

Ferrofluid slows down the head slightly.

Regarding the amount of helium or air as a coolant, the requirement is between 0.8 liters and 1 liter of gas per hour of head operation [which is very little]. A precise pressure reducer and a flow meter called a rotameter are required.

A helium-cooled head is practically indestructible if the current supplied to the drive coils is not exceeded.

All professional heads have a coolant inlet.

I even use cooling for the amateur heads I've built, so it's worth considering the channels and tunnels that supply the gas to the head.

[sameal]

This thread has me thinking about cpu liquid coolers..........they can get pretty small.........the weight addition would probably be awful

[farmersplow]

This thread has me thinking about cpu liquid coolers..........they can get pretty small.........the weight addition would probably be awful

Haha! Yes, that would be really something. With all those hoses and an external air conditioner ! Wonderful!

Unfortunately, it would only cool the cutting head and not the coil.

Small hoses would be an alternative. These could be connected to a can of freeze spray, and an electric valve (CPU-controlled depending on the coil temperature) would then allow -200°C to reach the coils – superconductors, so to speak. – Haha, I have to try that sometime!

[sameal]

[/quote]

Haha! Yes, that would be really something. With all those hoses and an external air conditioner ! Wonderful!

Unfortunately, it would only cool the cutting head and not the coil.

Small hoses would be an alternative. These could be connected to a can of freeze spray, and an electric valve (CPU-controlled depending on the coil temperature) would then allow -200°C to reach the coils – superconductors, so to speak. – Haha, I have to try that sometime!
[/quote]

I did see custom cooler blocks and i bet you could extend the hoses on a cpu cooler a little bit. This might work pretty good for the little speaker/diaphram type.

Coils on the other hand would be difficult. Some like the rca tombstone kind have the coil resting on a plate of some kind, which could possibly be made into a cooler block.

Then all of that led me to see a ferrofluid pump was made. Now that would be something!

[zdenek]

I like proven technical concepts that have proven their worth in practice. Someone came up with this idea a long time ago, when I wasn't around, and it's been working for decades to this day. That's why I read and analyze service manuals, and my thoughts and what I think would be good are a different matter...

[Lee2023]

Ok this is really excellent and it will be part of my next head build, but first I will cut some stuff and see how loud and long I can go with this setup, I have done quite a bit of glue research and ended up using the 1ml cyano lepage glue for my coils, go figure the ''same'' cyano glue in a bigger container with the supposedly same specs doesn't have the same properties when temperature is increased, around 4-5 days testing all sort of glues on a hot plate with different types of attachments, anyway I have added small 0402 smd NTC at the junction of my voice coil and my spider to measure approx where I am with temperature when cutting, added some aluminum rods and other spacer to direct a max of heat from the ''magnetic motor body'' to the external world, so in reality the only thing missing is helium flow since all the design have been made to conduct a maximum of heat away from the magnetic motor body, spiders etcc,

Yes I have some Hernon 360 stuff but the procedure to use this is quite complicated, you need to coat the magnet wire first bake it while going thru and oven, coat the coil former pre bake, then assembly while reactivating with MEK or other solvent, then bake the final assembly in an oven, this is quite a project, but will be sure try this one day since it's supposed to hold really high temp, so... my gap is 30 mils and I want to reduce this more to increase mag field in the gap and use N52 custom neo magnets, now using N45 neo magnet, I'm trying to reach the 2T limit... since reducing the turns to a max in the gap for minimum mass you need to compensate for the reduced BL force..

I'm currently working on a RMS to DC thermal converter around the idea of the LT1088, linear tech application note AN22 to be able to measure true RMS current, RMS voltage then calculated RMS Watts, injected energy, coil resistance... if it's not that useful in the end, it would have been very fun to design, waiting for the smallest 1N4148 diode in SMD and some 0402, 0604 resistors to upgrade my proto... only had some SMD SOT-323 double diode and 0805 resistors, still have some results with this rude setup, i'm not a DSP programmer... when you start evaluating the math behind RMS calculation registers are going to be big.... min 12-14 bits AD, you need to square this, sum this over a period of time etc etc... anyway this is the way I have decided to go, there is always other ways

I have done some tests and measurements with the TEC idea but time constants are long because low carbon steel and Neo magnets are not that good at conducting heat... and the hot side need to be quite low for a small TEC to be able to have some good delta and cooling the mag motor structure.... then the dew point, corrosion because of the untreated low carb steel, with keeping things not too big, the head needed king of 30 mins to reach it's thermal equilibrium, then you need a cooler etc... to complicated and no real advantage in the end I have found

Are you using using ''balloon grade'' helium ? Or the medical pure stuff ?

Thanks for the pages in the manual, always a good idea to check the manual... forgot about this !

Lee

[zdenek]

Are you using using ''balloon grade'' helium ? Or the medical pure stuff ?

Thanks for the pages in the manual, always a good idea to check the manual... forgot about this !

Lee
[/quote]
I use technical helium gas [like the one used for blowing up balloons] because it is relatively cheap, 8 liters of gas cost $200 and it is enough for 15-20 12-inch records.

[Lee2023]

Hi,

So I have found a nice place that sell high purity Helium with the regulator for around 250$CA including taxes, one more thing solved

The thermal RMS prototype if someone is interested

It will be less ugly and less burned in the end, i'm trying to find the limits before the solder melt or other things melts.... burned a few time, glue melting... trying to reuse a max of things to save time... this project is taking proportions bigger than expected, under 3.0Vrms things are holding with 47 Ohms 0805, in the end I will probably rise this to 100 ohms,

So, diode sandwiched inside two 0805 resistors, on the right one resistor for the RMS thermal measurement input floating so no ground loop, other resistor for the fast recovery when undershooting, on the left the ''feedback'' resistor that the loop is heating to match the temperature of the measurement input resistor, the other resistor is a dummy to match the thermal mass of the input sense assy, some 42AWG magnet wire, isolation, the main board is from an old obscure CyberResearch acquisition usb box, impossible to find the software so was able to use the balanced analog channels input, scanning mux and the 12 bits parallel adc, really nice analog adc front end, really stable no need to do some averages... one channel for RMS volt the other for RMS amps via the current transformer and another RMS thermal unit, response time is not too bad anyway the thermal integration time of the mechanical assembly of the head is not that fast also, finishing this clean, integration in a clean enclosure and going back to the final head test with feedback before cutting with the new head !

Lee

[zdenek]

It's always some kind of coil temperature control, better than no measurement at all.
In professional head protection systems, the measurement is made in real time using the bridge method.

At the same time:
1. Current measurement
2. Peak voltage measurement
3. Head drive coil temperature measurement
4. Power measurement as a result of multiplying the current and peak voltage
5. Head excitation protection
6. Head protection against the appearance of DC current and DC voltage at the head terminals [disconnecting the head if DC voltage appears at the amplifier output].
7. Delayed head activation and disconnection in the event of a power outage.
All these components are located on two boards: the left and right channels. The head is powered by 4 amplifiers, two reverse-phase power amplifiers per channel. The power difference between the reference resistor and the resistance value of the head coil [bridge resistance measurement in the real-time circuit of the head operation] is amplified and fed to the execution blocks and to the relays that disconnect or connect the amplifiers' power outputs.

[Lee2023]

Hi,

Yesss, I use a yamaha P2100 power amp, so no relays or DC protect, I have a startup controlled sequence for the power amp, audio cut, speaker out into dummy load for initial thump, wait, then I think audio to amplifier, speaker out to head, inverse kind of sequence when cutting the amp,

This is the bloc I need to install for DC and power lost protect I think implemented not sure, this thing was build almost a year ago, the thing is that it's not easy to cut large amplitude DC, if drivers shorts to supply or other bug, when you verify the ratings of relays... DC arcing is complicated to fill vs an AC arc... ESP as a really good article on this, it's not easy at it seems... again haha

I will sure have a temperature audio cut system added, but for now I need to cut to see what kind of torture the head will be able to endure, as said I have a very small smd NTC integrated in my assembly near the coil so it will be ok with some margin to monitor temp, I know what is the max working temp of my glue so.. the RMS stuff is to give me an idea of where i'm going and it was fun to work on something else, head design non stop kind of driving crazy so I need to take breaks, when you are in a hurry or not really wanting to work on this you can destroy months of work and $$ things very fast...

Have read a bit on the bridge kind of stuff a long time ago when started on this adventure, need to recheck this when I will have time, or more time to implemented more advanced features on the setup

[farmersplow]

I like the way you approach things. Very scientific, experimental, and precise! I am very curious to see the further results of your research trip.

[Lee2023]

Hi !

Updates of this out of control project that I need to stop working on haha

So finally reaching the end of this measuring acquiring thing, decided to push it to the max, so cheap LCD character module, assembly code PIC16F877, computer communication dumping data via ttl / usb serial port to txt file, then I can process for what I need to do, this will help me to characterize my temperature vs RMS input power, I will then be able to run a wav file analysis to see if temp will be ok vs level before doing a cut, will also be usefull to have a feeling of what the system can give

The PCB are with DHL and will be here Monday, I decided to use beautiful AD542HK opamps, super performance, saved from the recycling bin, the cheap audio moduled are crappy amplifiers fake TDA2030 ? turned to DC amplifier low bandwith to drive the RMS current measuring thermal resistor...

I was not able to get this thing working ok with 1M4148 SMD, kind of acting more like a noise generator than something stable, strangely the first diode used SOT23 where the best, in the end I have used MMBD7000HS-7-F diodes, working great, measuring errors because resistors and diodes are not matched will be compensated with code, just need to do a DC calibration and add some factors, response time is a bit slow at integrating RMS but my head assembly temperature time constant is not lightning fast anyway, it will do a go job !

[Lee2023]

Hi,

So conclusion and final results, 2000 lines of code later, working great at some extend for what I need, pink noise applied so you can see volts RMS, Current RMS, Watts RMS, Temperature measured near the junction of the voice coil former and alu spider NTC 3950 SMD 0402, total energy, each sample is approx 0.215 mSec, sampling rate approx 4 Hz well enough vs Nyquist frequency, response of the thermal system

Overtemp is programmed and requested as 2 bytes at startup via RS232 or you can use the default value stored in EEPROM by pushing the multifunction button, value stored in memory is the last uploaded alarm value via RS-232 so I don't need to hookup a computer after setting the value.. alarm is a decimal value corresponding to the ADC output, so NTC have negative coefficient of resistance it's a ''under resistance'' alarm

I'm now able to approximate the thermal RC equivalent model for my system and simulate temperature rise vs input material to see if I will be under my max glue holding temp...

So, now what is fun, you can see the increase in resistance of the voice coil since temperature is going up, decrease in current resistance going up and increase in voltage since load is going up, so things are working, now time to install this and final amplifier DC protect & driver alarm kill and start cutting

[zdenek]

Make corrections for the increase in coil resistance with temperature.
The output characteristic is not linear.
According to my measurements, for example,
coil resistance is 8 ohms, room temperature 25 degrees Celsius, then
coil temperature 100 degrees Celsius, the coil resistance only increased to 9.2 ohms.
You should enter these corrections into the program.
The SMD thermistor sensor will show you a linear characteristic (almost linear).
150 degrees, the temperature of the drive coil winding, the DC resistance was only 9.8 Ohms, at two hundred degrees the coil resistance was slightly above 10 Ohms, and the specific resistance was 8 Ohms for 25 degrees Celsius.
It is best to do the live scaling with the thermistor connected close to the coil, but you should pre-scale.

[Lee2023]

Mmm everything is good,

So

6.946 Ohms and 24 deg C at sample #512
7.126 Ohms and 30.21 deg C at sample #6389

So delta temp is 6.21 deg C

R=Rref*(1+alpha(T-Tref)))

Rref is 6.946 ohms
Alpha for copper 0.00393
T=30.21 deg
Tref=24 deg

6.946*(1-0.00393*(30.21-24))=7.116 ohms close enough to measured 7.126 ohms

[zdenek]

This means that the measurements are relatively consistent.

[Lee2023]

Yesss, I wasn't expecting that much precision from this thing