Testing of the Batchelder - Grampian replica

[EpicenterBryan]

If I was selected to test this head I would do it like that:

Thanks for the inputs!

To one of your ends, I've installed a brass diameter adapter to convert the 33rpm position on my lathe to a 45rpm speed. This is not going to be reversible since my 33rpm shaft is messed up so it ends up being a press fit. No matter, since John is going to make me a new spindle adapter and will be 33 and 45 rather than 33 and 78 anyway.

IMG_4186[1].JPG

Now for the bad news... and newbies or cheapskates like me should pay attention. Some times it's not worth using the last few minutes on a disc when messing around. Tonight, I chipped my one and only short shank stylus. John sent me a new long shank with heater wires when he first sent his head, but since the Grampian specifies a short shank, I've not been using that one but a random used one I had that was is good shape. My Presto 1D uses long shanks, so this was the only one I had for unknown reasons... So anyway, I got distracted and let the cut continue too far to the center of the disc and it did a locked groove down to the aluminum. Stylus toast. Not sure if you can see this in the microscope shot but it's messed up.

IMG_4189[1].JPG

John has 2 new short shanks on order from Apollo. We all know how that process goes. They don't answer the phone - you send in a request for quote - they respond via E-mail - blah, blah. We are not sure if they will ship directly to me, or only to John first then he has to send to me. Time lost but it is what it is.

One other thing I should clarify: The cuts I'm doing now with reference level and Pink noise are with no EQ - No Inverse RIAA Encode. They are straight out of my ProTools TDM system. Playback is also with no RIAA decode. Totally flat. I'll go into great detail on the setup when I post results.

Bryan

[Stevie342000]

John has 2 new short shanks on order from Apollo. We all know how that process goes. They don't answer the phone - you send in a request for quote - they respond via E-mail - blah, blah. We are not sure if they will ship directly to me, or only to John first then he has to send to me. Time lost but it is what it is.
Bryan

Just a heads up here whilst we may wish to support Apollo with lacquers we have other sources for cutting styli.

I am not sure if Todd is making short shank styli or is up and running with that side of his efforts, but there is the Shank, he may be in Europe (France) but he does have short shank diamonds, which he will resharpen too, you can do a trade-in on chipped or damaged styli.

Just a thought, as this piece of string could become very long waiting for Apollo and we are all keen to know the end results.

In the UK we seem to have gone for Short Shank styli, the Sugden (Connoisseur) uses short shank as well. Shank length was mentioned in one of the BBC reports, when they were testing Presto equipment or in developing the Grampian head, don't remember sorry. The BBC favoured short shank styli.

[markrob]

Hi,

Its easy to make a long shank short. Just cut the shank with a pair of nippers. Its just aluminum and cuts easily.

Mark

[EpicenterBryan]

Just a thought, as this piece of string could become very long waiting for Apollo and we are all keen to know the end results..

I'm with you Stevie. I like Marks idea about modifying a long shank, at-least so I can hear what it sounds like with good level. However, I'll wait for the correct stylus to do the final frequency measurement and audio test before publishing anything.

Bryan

[EpicenterBryan]

Just cut the shank with a pair of nippers.

That worked great! Aligned the tip of the long with the short, and cut off the excess. Then sanded the edge with 400 grit. Exactly the same length! and the flat sides are the same length as well. I'm back in business!

Bryan

[Stevie342000]

Just cut the shank with a pair of nippers.

That worked great! Aligned the tip of the long with the short, and cut off the excess. Then sanded the edge with 400 grit. Exactly the same length! and the flat sides are the same length as well. I'm back in business!

Bryan

That's good when you get the new styli from Apollo you can do a comparison against the snipped stylus data.

I know it may not be in your remit but what about a steel styli from Jerry? Just thinking we can cover more than one base here.

short shank, snipped long shank, steel styli (although they may interact) and a diamond. That way we know how it performs with all options. I know like you not got enough to do already.

[dubcutter89]

....another thing that might be interesting is the sensitivity of the heads. If you could measure the AC Volts/Amperes it takes to cut a 1kHz 0dB sine + the information on resistance/inductance we could compare the old alnico vs. modern rare earth magnet version...
It is not that power amps with 100000Watts are a problem or expensive today, but a coil can only take a limited amount of energy before it will blow so sensitivity is still interesting..

Looking forward!
Lukas

[EpicenterBryan]

Here is the update.

Testing is complete on the Batchelder Serial Number 001. Comparison testing with an original "gold standard" Grampian is also complete.

The short version regarding Serial Number 001 is that it takes more drive signal to obtain a 5cm/s lateral rms cut than the Grampian. It also has distortion issues as you will hear in the test cuts. I beleive John can address these issues.

John is winding a new coil for what we will call serial number 002 to closer match the inductance and impedance of the Grampian coil. I'll be sending back the second Grampian and he will also compare the torsion rods to make sure his have the same stiffness, and he will double check the gap between the pole pieces and the armature.

Now I should go into the equipment used in testing.

General:
All cuts were at 45rpm.
Not using feedback.
Lathe: Presto 8D, non heated short shank stylys.
Cartridge: Shure M97xe, 4.0mv @ 5cm/s peak 1Khz, 1.25g tracking force
Preamp: Wayne's Flat MM preamp.
I/O: Protools TDM system with 888/24 I/O.
Meters: Data Precision 3400R (volts), Fluke 8050A (amps), Teac MB20 VU bridge.
S/W:
ProTools 6.4.1
Audacity
Marks Preso iRiaa VST plug-in
Cool Edit Pro
------------------------------

Grampian Testing:
Specified Impedance: 16 ohm
Power needed for 7cm/s peak rms: 1.26W or 31dbM
*** 31dbm = 4.49V rms drive ***

Test levels set for 1Khz:
Voltage: 4.46vrms
Current: 0.236A

Disc playback Level on VU meter:
1Khz 0db, White noise -4db

Noise cut Diameter: 9 inches
Music cut diameter: 8 inches
------------------------------


John head testing:
Test levels set for 1Khz:
Voltage: 5.991vrms
Current: 0.675A


Playback on VU meter:
1Khz +1db, White noise -2db

Noise cut diameter: 13 inches
Music cut diameter: 12 inches
------------------------------

The 1Khz and White noise tests were cut with no EQ in the path. Playback was Flat using Wayne's flat MM preamp.

The music was pre-processed with Mark's Presto iRiaa VST plugin using Audacity. I'll go into more detail about why later. Music Playback was Flat, again using Wayne's flat preamp. Riaa was applied in Audacity prior to posting music for your listening pleasure.

About the Music:

Initially, I was going to do flat disc to disc copies of some actual mono mixes from 45's. Once I had frequency responses in hand it became obvious I couldn't go that path.

I described to Mark what I was seeing and he said "that sounds like the way a Presto 1D works". It turns out, it is very similar. There is a Natural roll off in the low frequencies built into the heads. Applying standard iRiaa, or using music from a 45 that already had it applied would attenuate the low frequencies twice!

Presto 1D.jpg

Mark mentioned he had a VST plug in that did iRiaa except doesn't have the 500hz turn over. It still has the 50hz, and 2122hz points. Just what I needed!

Mark's VST.jpg

Then I planned to use some Creative Commons free use music. The selections I chose turned out to be in MP3 format, and I used Audacity to convert to 24bit, 48Khz. Then bounced the files in Protools to Mono. It turns out those source MP3 files had clipping and I want to clue people into a little irritation with Audacity.

Audacity Clipping.jpg

On the left Audacity shows the clipping, on the right it does not. By default, Audacity doesn't display clipping. So if you have done a fresh install (like I did a while back on my Protools system) it won't show you that a file is clipped unless you go in and select the option. My desktop had it turned on. My copy on the Protools system didn't. So be warned.

Rather than mess with de-clipping and all that, I asked my friend John Shipe if I could use some of his music. I selected a tune from a double CD we recorded together a dozen years ago...

dualcd.jpg

Pa Moom, from the CD Pollyanna Loves Cassandra by John Shipe Band.
Involusion Records, 2003.
Used by Premission.
http://www.involushun.com/pollyannalovescassandra2003.page?cart=128836309410445347

This was a stereo mix that I just bounced to mono.

Now about the tone and white noise. Those were generated by Protools, recorded and then Cool Edit Pro was used to measure the average rms value. Then levels were matched.

CONTINUED ON NEXT POST

[EpicenterBryan]

Alright, now for some plots. These come from Mark who has been a massive help in wading through the details. He also did a great write up on his observations which he has given me permission to post here.

Notes Mark Robinson 12/13/15
There are 4 screenshots from Adobe Audition taken of FFT spectrum analysis comparing the Batchelder head to the Grampian.

The graphs show the distortion products and frequency response to white noise of each head. Levels were normalized to 1 KHz sine wave reference tone recorded such that a 5 cm/sec rms lateral playback level was achieved. The white noise samples were cut without any RIAA on encode or decode and represents the open loop velocity response of the heads.

Distortion
The Batchelder head shows much higher distortion levels for odd and even harmonics. The distortion level of the 3rd harmonic is about 5 % vs. < 1% for the Grampian. It was noted that the Batchelder head required 6 Vrms at 1Khz to reach 5 cm/sec vs. the Grampian which needed only 4.49 Vrms to achieve the same level. The distortion of the Batchelder head is very evident when listing to the playback. The playback of the cut appears to be nearly 4.8 db lower than the Grampian when cut under the same conditions. It seems that saturation of the head is occurring and thus limiting the maximum cutting level that can be achieved with this head.

Frequency Response
The Batchelder head has a response that falls off at a rate of about -6db/oct below 3 Khz. Above this frequency, it falls off gradually up to 20 KHz. There is a secondary resonance that appears in the response plot in the 15 KHz range. This also appears in the Grampian response plot and is probably due to the cutting stylus shank. More work needs to be done to determine if this is conclusion is correct.

It was noted during preliminary testing that the Batchelder head had less measured inductance as compared with the Grampian. This could explain the higher frequency of the transition from constant amplitude to constant velocity (see below)
The Grampian head has a response that falls of at a rate of about -6db/oct below 300 Hz. This is to be expected due to the R/L time constant of the head and represents a transition from constant amplitude to constant velocity response. Above this frequency, the response remains flat up to 10 Khz, where it falls off rapidly. The same secondary resonance is present as seen in the Grampian head.

This response is not expected for this head and may be due to age, damping fluid, cutting stylus, or playback issues. Work needs to be done to identify the cause(s).


Conclusions
Overall, the Batchelder head needs some additional modification before it can be offered for sale. I would recommend checking the electrical characteristics of the winding to understand the reason for the large measured difference in inductance. I would also verify that the un-damped resonance of the head be tested to make sure it is in agreement with the Grampian sample. The mass and stiffness of the armature should be verified. It was noted that the pole pieces on the Batchelder head were not laminated. The need for this should be verified. Finally, the size of the gap between the armature and pole pieces should be checked and verified against the Grampian.

John's Head Response:

John Freq Response.jpg

John's Distortion:

John Head Distortion.jpg

Grampian Response:

Grampian Freq Response.jpg

Grampian Distortion:

Grampian Distortion.jpg

Now for the music!
Note that the playback from John head stops abruptly. My Protools system locked up while cutting. But you get the idea.

Source File:
https://soundcloud.com/epicenterbryan/pa-moom-mono/s-1bHEy

Johns Head Playback:
https://soundcloud.com/epicenterbryan/john-playback-riaa/s-vsaV8

Grampian Playback:
https://soundcloud.com/epicenterbryan/grampian-playback-riaa/s-v3mjT

So, let's see how Serial number 002 shakes out.
Until then, that's all folks.

Bryan

[inspector77]

Many thanks for your excellent explanation , to be the first version sounds great!!!

[Mulchefye]

the grampian sounds better in my opinion, seems the first one has a bit of distortion going on

[EpicenterBryan]

A quick update for tonight.
I have tested mods to serial number 1 which I will call Serial #1 (V2) since it's not an entirely new head but has been modified to explore a few things we wanted to test. When I say "we", I mean Mark, John and me.

Damn - John is on to the solution. A few coil detail changes, and in this prototype unit a big change - changing the pole pieces from solid to laminated. This prototype has perhaps 15 lamination's - the Grampian has over a hundred per pole piece.

The S/N1 V2 unit sensitivity test is now right on par with the Grampian gold standard. The natural LF roll off has also moved far to the left as we hoped it would.

Much more on this ... this weekend.
Oh, and I did cut music with S/N1 - V2 and it sounds Bitching!

Bryan

[ROLANDJAYS]

Oh yea
I got my finger on the pulse now .
I didn't have the right materials on the first head .
Also for everyone out there making a head , the materials you choose make a huge difference!!!
I cannot tell you enough , anyone can make a head that sounds like shit , but the Fine details are what make it sound good and efficient.
The next revision should be the final.

[opcode66]

Good design is as much a factor as is proper materials.

[EpicenterBryan]

Hey guys,
It's time to post some results from the S/N 001 V2 test bed of the of the Batchelder - Grampian replica.

First off, Let me give you some background about what has been going on.

When I had all the heads here, I took resistance and inductance measurements on them all (including feedback if installed). John's original head inductance numbers didn't match the Grampian. In the case of Drive inductance, it was low by a factor of 10 using my meter.

Mark and I discussed this and we tried to come up with some things John could do on his end to gather some data while he was re-working things. John bought an LRC meter like I have to make the same measurements. I sent all the heads back to John and he disassembled his head and the Grampian that had a bad feedback coil. We had John do several tests to help determine what was going on. Here are the tests performed and why:

1). Remove a Grampian coil, and remove John's coil and measure the inductance of each outside the heads in free air.

It turned out John's coil did not have the same inductance in free air as the Grampian coil although it did have the same resistance. That implied it had the same total length of copper wire, but not the same number of coil winds. John worked on a new winding jig and made some new coils until he hit the Grampian numbers by testing in air.

2). Install the new coil into the magnetic structure of an actual Grampian, and also into John's head structure and measure the inductance.

When John's new coil was installed in the Grampian laminated pole piece structure, the measurement was within 10% of what we expected. When John installed the exact same coil into his magnet and pole piece structure, the numbers were way off. This implied that the solid pole pieces vs. the laminated (and different material) pole pieces were at the root of the problem.

3). Make a quick approximation of the laminated Grampian structure and install the coil for further testing.

I recommended to John to take apart some audio transformers to obtain some material, specifically a little transformer he could get at Radio shack for a few bucks. The lamination pieces were just too small for him to use. So John took apart a bathroom AC fan which had material large enough to salvage and hand made a pair of laminated pole pieces. The material was WAY thicker than what is in the Grampian - but he was able to do it. Instead of having over 100 lamination's per pole, the V2 test head has only about 15 per pole. But it was enough for us to see a rather drastic change - and to point changes in the right direction.

I took some material I salvaged from a small AC transformer to my friend Andrew Sztymelski, who owns ABC tool and Die Company and he looked the material over. In the case of the Grampian, the material is even thinner. The material is very pliable which means the tolerances need to be very tight to make a punch for his turret press to make thousands of the pieces. The cost of a punch set would be in the $7,000 price range. We talked about other options like EDM where we could cut a stack of lamination's up to about 4" thick at much less expense per piece but there are other issues. John is working other options on his end that may work out and may cost less for fewer units.

My point in sharing crazy details like this is just to let people know that there are large expenses involved in trying to get the last 10% of performance out of anything.

Having said that, I'm going to share how the V2 test bed worked out.
The performance is remarkably better. And John is on track to bump it up another notch. The cost is unknown, and that is something he will need to think about.

Thanks again to Mark for taking the time to analyze the files and provide the following performance graphs. Below these graphs I will also post his comments. I also want to note that this time we have responses both with dampening fluid and without damping fluid. For the responses with fluid it's 100 grade.

Also note that John said in this crude V2 test bed the tolerances are much wider, and they were not the same right and left because of the quick experimental nature.

Here goes!

Response with NO damping fluid:

John Freq Response Ver2 No Fluid.jpg

Distortion with NO damping fluid:

John Head Distortion Ver2 No Fluid.jpg

Response with 100 damping fluid:

John Freq Response Ver2 With Fluid.jpg

Distortion with 100 damping fluid:

John Head Distortion Ver2 With Fluid.jpg

CONTINUED ON NEXT PAGE:

[EpicenterBryan]

Now for the Audio playback from the S/N 001, V2 test bed of the Batchelder - Grampian replica:

This used the same reference audio source previously posted. Here is the audio from this new test of the of the V2 test bed version of the Batchelder - Grampian replica:

https://soundcloud.com/epicenterbryan/music-10vuplriaa-norm

Here are Mark's comments about the response of this V2 test:

Hi Bryan,

For the no fluid case, I calculate distortion products of 3.2% 2nd harmonic and 2.2% 3rd harmonic.

For the fluid case, I calculate distortion products of 3.2% 2nd harmonic and 1.8% 3rd harmonic.

2nd (and even) harmonic distortion indicates non-symmetrical behavior like you would get from a class A amplifier (positive going portion of the waveform differs from the negative portion). 3rd (and odd) harmonic distortion behavior indicates symmetrical non-linearity like you would get from a push-pull amplifier (an “S” shaped transfer function).

The system resonance seems to be in the right ballpark at about 7.5Khz (10 Khz would be closer to the expected number). The damping fluid does not seem to be able to completely kill the resonance. Maybe a high viscosity fluid would help.

The overall frequency response seems much closer to the Grampian. The high frequency response might be improved with a shorter shank stylus and better damping. The low frequency turnover is at about 500 Hz, while the Grampian was at 300 Hz. Still very usable and much better than the previous version.

The music sample seems much better to me as well.

Mark

Mark also noted the following based on a photo I took of the stylus as mounted in this testing. I raised the question because I wasn't sure it was fully inserted since John was not sure the part in this head had it's tolerances opened up for stylus diameters as we have discussed before. I used the same stylus I used last time that I shortened. I inserted it as far as it would go, and yes, the set screw did engage on the shank.

IMG_4311[1].JPG

Mark commented as follows:

Hi Bryan,
Got the files. I’ll be back with comments later. I looked at the pict of the head with the stylus you sent. The length seems long to me. I’ve attached a pict I grabbed from the net long ago that shows a stylus attached. See what you think.

He sent this photo for reference:

Grampion 1.jpg

I'm posting these photos because there is a slight possibility that the stylus I used may be slightly longer than it should be. Longer gets more output at the sacrifice of HF. It looks close to me but I'm thrilled to have comments.

In testing of the final version of John's head, I'll pay special attention to details like this...

Bryan

[Sillitoe]

Hey Bryan, the soundcloud link isn't working.

[ROLANDJAYS]

it doesn't work for internet explorer, but I think it does for google.

[Sillitoe]

Oh ok, I can't get it on Firefox or my phone, just downloading Chrome now.
Cheers

Edit- Na still not working for me...

[dubcutter89]

Cannot hear that second test files either....

But from what I'm reading progress should be good!

It turned out John's coil did not have the same inductance in free air as the Grampian coil although it did have the same resistance. That implied it had the same total length of copper wire, but not the same number of coil winds.

Well, if it is the same gauge wire and the same length (=resistance only if gauge is right..), same diameter (number of turns if gauge is...) it should (must) be the same inductance! Resistnace only implies length if it is the same wire...
So the coil was probably not like the original...

John worked on a new winding jig and made some new coils until he hit the Grampian numbers by testing in air.

But now it should be good
Perfect!

In theory the best coil for a moving iron head has lowest possible resistance and max. inductance! Thogether with a very high resonance this should make a head that is flat vs velocity - that is what you normally want..

Looking at the old numbers of version 1 the head sensitivity was also low compared to the original
0,675Amp vs 0,236Amp , that's almost 3 times the current consumption of the reference head. Assuming a coil with the same number of turns this would be 3 times the magnetic flux (linear scenario) or magnetic saturation = distortion...
Maybe the magnets are a little weak as well...

The numbers for version 2 would be interesting!

About the curves:
Distortion is still a bit high, but if it sounds good... Would be interesting to hear the audio!
The turnover at 500Hz as it is now would be perfect to match the RIAA specs. Just add the 50Hz, 2120Hz turnover with electric filters and done!
Also the response is quite flat. An eq with a little dip at 7-8 kHz and everithing is fine.

The turnover of the reference (Grampian) is lower and therefore the playback with regular equipment sounds bass heavy / boomy...

The sound from version 1 was in my opinion already quite good, but the distortion was to high.

KEEP ON ROCKIN!

Best
Lukas