Hacking The Teenage Engineering PO-80

[markrob]

Hi Everybody,

When the PO-80 came out a few weeks ago, I decided to grab one and see if it could be hacked to get better performance. I put the kit together and played with it for a bit and then went to work upgrading it. Rather than be stuck with buying blanks and cutting styli from Teenage Engineering, I decided that I would use it to emboss CD-R disks. To do this, I just turned the stylus backwards. I traced out the schematic for the unit and decided that although its a pretty nice design, the 5 watt amp did not provide enough headroom to really drive the head to EQ the highs. So, I rolled my own 3 Watt amp that has reserve power to push 100 watt short duration peaks to the head at 4 ohms. The supplied head design is based on a exciter speaker just like many of our DIY heads here. Rather than risk blowing it up during my experiments, I decided to roll my own based on a Datyon DAEX13CT-4. That way I can replace it if it pops. I was also concerned that the silicone tubing used to couple the stylus to the driver might lose efficiency at high frequencies. On my head design, I hard couple it to the stylus shaft. Here is a link to a short Fusion 360 animation of the head construction:

https://drive.google.com/file/d/1KYOLhktLab4Des2BRRzfQJRJnS3qnguB/view?usp=sharing

I modified my custom PC based DSP software suite I wrote my lathe for this machine. It allows for head EQ and calibrated power monitoring along with steep high and low cut filters plus some fast limiting.

See the link for a short video of the embossing process where you can see the DIY head, amp, and software running embossing a track.

https://drive.google.com/file/d/1THEBQbnJzlgY2P42ejTyOYmsjhbnvfVk/view?usp=sharing

Finally, here is a link to a playback of the embossing:

https://drive.google.com/file/d/13HqjmsrjrlCVd9b84Wzt5gx7GFBnkihr/view?usp=sharing

And the source file used:

https://drive.google.com/file/d/1V8RNJgP0GVv1AYb4C9Q687tQrRGR8cie/view?usp=sharing

All in all, I'm pretty happy with the results. You can hear that the drive motor suffers from quite a bit of wow and flutter and the playback is a bit fast due to the cutting speed being slowed due the mechanical load of the embossing head weight.

If there is any interest, I can provide more details of the mods. There are still more possibilities here.

Mark

Mark

[mrd]

Hi Mark

That's really awesome work, thanks for sharing! I've been planning a dayton exciter replacement on the gakken for a while but other things have got in the way - are you happy to share STLs? (or whatever format of this)


Maybe good to add pitch mod for the head movement too so locks are (more) possible - pics are on the gakken thread of mine and heres a vid: https://www.instagram.com/p/CP3KXQqHsNv/

Cheers
Dylan

[markrob]

Hi,

Here you go. I can also provide Fusion project files if you think you could use them.

Driver STl Files.zip

[CutR]

Great work Mark,

You can also emboss without reversing the stylus by just backing off the weight and dialing in the cutter head spring. You can also avoid some speed issues by not having the clamping screw, screwed down to tight.
Regarding wow and flutter, someone posted in the gakken thread about using white lithium grease on the cutter head track and the reduction box gear. I have tried this and it does make a noticeable difference in reducing
wow and flutter. I'm also looking at running a bigger amp like a 100W, what sort of fuse would i need to protect the head from blowing?

Also do you have some close up pics of your cutter head mods.

Regards
CutE

[PLD]

That's really rather impressive...!

[markrob]

Hi,

I measured the wow and flutter with just the playback arm and its pretty bad. Here is a screen shot.

Wow and Flutter.png

You can see the unweighted comes in around .6% From the scope screen, you can see there is a major 25 Hz component that corresponds to 1500 RPM, which is about the motor shaft speed at 45 RPM. I'm sure it gets worse under load. The no load speed is pretty close at 996 Hz vs the 1000 Hz of my test record. I know it slows down more under load. I've been thinking about a replacement of the exiting motor with a true servo motor (I have a low cost one in mind), but it may be too much for a simple toy. Perhaps adding a heavy plate to add more inertia would work to filter some of the 25 Hz ripple. I'm afraid of the extra load on the platter bearings.

I don't see how adding clamping force would affect the wow and flutter.

I have not tried to modify the existing cutter head. This is a "from the ground up" re-design using an off the shelf driver and 3D printed parts. The STL flies and animation should give you enough to go on, but I can provide more info if you need it. I might go back and see what kind of performance I can wring out of the stock head since I have a working replacement if I smoke it. I tried to locate the spec for the driver they used, but on a quick search, I was not able to find it. That prompted me to do my own.

You do have to be careful with a big amp driving the head. One accident and the head will be toast. Its hard to fuse against this, but the head will need 1 amp at 3 ohms for 3 watts. I'd start with a 1/4 to 1/2 watt fast blow and see how hard you can push it. My test cut was running about 2 Watts average with 100 watt peaks. I designed my amp to have have low continuous power rating with enough energy storage to hit 100 watt peaks for a short time. Most commercial amps are not designed this way and could easily push full power long enough to fry the head. Good metering is also very helpful.

Hope that makes sense.

Mark

[markrob]

Hi Everybody,

Here is a link to the Fusion 360 Project for the head.

https://a360.co/3gvbmgg

Mark

[CutR]

Thanks for posting that link Mark.

When i was talking about the clamping screw, i was referring to the playback speed not wow and flutter. Have you also measured it on a normal turntable? When i playback test cuts on my SL1200 TT wow and flutter are slightly reduced comparing playing on the gakken, To me it seems like the gakken playback arm kinda accentuates the wow and flutter a bit.

On a side note, have you managed to cut any graphics on your gakken using that image2groove software you created?

Regards
CutR

[markrob]

Hi,

The wow and flutter should be somewhat additive if you record and playback on the same system. I expect that the wow and flutter to be worse during embossing. If I get time over the weekend, I'll measure the wow and flutter when embossing with playback on a decent turntable.

I have not tried to make any graphics cuts with this machine.

Mark

[mrd]

Hi,

Here you go. I can also provide Fusion project files if you think you could use them.



Driver STl Files.zip

Amazing, thanks!

[markrob]

Hi,

I've attached a PDF of the schematic that I traced of the circuit board for the PO-80. It's a pretty nice design. A boost converter to bump the 5Vdc up tp 12Vdc to run the platter motor. I class D 5 W audio amp and a preamp for the ceramic pickup. One thing I noticed is that they load the pickup with only 220K ohms. This will really kill the bass and I'm sure they did this on purpose. I want to try increasing to 1 meg and see if I get better performance via the output jack. Might just amplify all the rumble present. Also uploaded spec sheets for the amp, dc-dc converter, and motor.

PO -80 Record Factory Amp Schematic.pdf

MT3608.pdf

CS8138.pdf

EG-530AD-6B-ETC.pdf

[ton]

Hey, I have a gakken, I imagine the circuit is identical. I'd like to try some of these mods out. I'm pretty comfy soldering and I can obtain 3d printer bits and any components if you have a BOM, but admittedly I'm real new at this. But I'm down to let the magic smoke out if that's how it goes. Maybe if you can provide some pics of where/how you implemented any changes to the circuit. Any specifics about your DAEX13CT-4 cutter head and how to make that happen, any specifics about amp.

[markrob]

Hi,

So far, the only mod I made to the board is to increase the resistive load (R9) on the ceramic pickup from 220K to 1Meg. That gets the bass response back. I have found that my cheap USB supply has quite a bit of 60 hz ripple and it becomes audible once you open up the lows. When I ran the board from a clean DC bench supply, the hum went away. I guess that's why they cut the lows so much.

I'm working on a motor upgrade using a low cost NEMA 14 stepper to try and solve the speed regulation and wow and flutter issues. If I get that running, I plan to design and layout a single PCB that has the playback preamp, head drive amp, and motor control. So stay tuned for an update.

To build the head, you just need to print the two parts I provided STL files for and get some DAEX13CT4 drivers. Here in the states, that is from Parts Express. They were on back order, but I think they are back in stock. You don't really need my power amp design to run the head. Any amp that can drive 50 watt or more into 4 ohms will work fine. You will need some heavy EQ to get the performance I am getting. I could make my home brew software available or you could you could use something like Fab Filter to do the same thing.

Mark

[ton]

STL files printed. Dayton exciters acquired. Amp sourced. How did you affix the exciter to the frame piece? And how did you affix the ring to hold the back?

[markrob]

Hi,

I used a gap filling CA (super)glue. But other glues might be better (e.g. JB Weld or other epoxy). I glued the stylus holder to the driver and aligned it with the driver after mounting it then finally I glued the stylus shank to the holder after everything set. Make sure you have the stylus in the correct orientation before that last step (backwards to emboss). I figured the driver is cheap enough that I would replace it when changing the stylus. You could probably come up with a screw or other removable clamp to avoid this. Let me know if the printed files worked out for you as is or if you think they need a bit of tweaking. If you have access to Fusion 360, I can provide the design files.

Hope that makes sense.

Mark

[ton]

Thanks Mark it does make sense. The parts all fit beautifully thank you for the stl files! Would you mind sharing a couple pics or a vid showing the back of the head with note of where things are affixed so I can be sure I’m attaching it all similarly?

[markrob]

Hi,

Did that rotatable 3D model I posted a link to not work or was it confusing or need more clarification?

Mark

[ton]

The model is helpful but doesn’t really illuminate how you’re fastening it in the real world. Something I could probably guess but don’t want to chance screwing it up.. you’ve glued it together in some places, should the ring be fixed to the exciter then the ring subsequently into the cylindrical frame? Should the ring be flush with the top of the cylinder?

[markrob]

Hi,

See the attached photo. I first attach the ring to the driver. This acts as an adapter as the driver does not have any mounting holes. After the glue sets, attach this assembly to the frame. There is a gap in the frame to allow the driver terminals to be accessed. The driver ring should be flush with the ends of the tabs in the frame. Let that final assembly glue set and then you should be good to attach it to the arm via the 4 screws used to hold the original. The hole in stylus holder will need to be drilled to the size of the stylus as it probably won't print at the proper size due to its print orientation. I used a 1.5mm drill. I attached the stylus holder (no glue yet) to the stylus and glued it to the driver in place to make sure the alignment was correct.

Head Rear.JPG

I've also attached a copy of the software I wrote for cutting and calibration that you might find useful (PC only). There is no install program just run the two exe files as is. It has my current EQ for my head and should be in the ballpark for your copy. I wrote this for me, and as a result have not any documentation. But here is a brief rundown of the cutting software.

Disc Cutting System.zip

I provide for accurate peak and average power monitoring. To make this work, you need to select your amplifier power rating from the drop down box. Make sure you start with your amp gain down to avoid overdriving the head. Connect a voltmeter across the head terminals turn on the 1 watt ref generator slide switch. This will produce a calibrated 1Khz sine wave. You should set your amp gain for a reading of 2 Vac RMS at the head, which equates to 1 watt of power. Now when you playback, the meters will indicate peak and average head power. The function select should be set to Normal Cut and will include the graphic EQ in the signal path along with some internal pre-EQ. You can load in a file to be cut and adjust for level prior to cutting. Once you set the calibration, do not change any external gain or volume settings on your amp or pre-amp. Only use the two gain controls (start with them at min to avoid overdrive) in the app or you will lose calibration. I adjust such that the peak power reading just hits max and keep an eye on the average power to make sure it does not get much above a watt or two. You can push the level a bit if you engage the limiter. Shoot for a couple of dB of gain reduction on the peaks. There are also high and low filters that can be used to increase level at the expense of bandwidth. The high pass is 4th order and the low pass is second order. You can save and recall the EQ settings as presets.

The calibration program allows you to do a test cut of a short reference program. Capture the playback and compare the two. You can see the spectrum and hear difference and use the EQ to remove the differences. Then once you have decent EQ, you can save and import back into the cutting system as your head EQ. To do this, you need to select the Test Cut option from the drop down box in the cutting system. This removes the graphic EQ but leaves the internal pre-eq in place. I prepare a short 30-60 second reference file that is used to cut the test. When you capture the playback of this cut, trim it to length and if needed, correct for any speed errors in software so that the two files are the same length in time. You can A/B the two cuts during playback and adjust for any gain differences to match loudness. If you do a good job of trimming the playback, you can loop both samples and they will stay in sync. By using program material, you can avoid piling on high frequency EQ that is only affecting noise.

If you have questions on the use, let me know and I'll try to help.

Mark

[ton]

I ran a 50W peak single channel amp into the Dayton exciter having glued the parts in place and using the self adhesive on the exciter to hold the stylus holder piece on. Stylus fit really snug through the little hole so just went with that friction to hold it in place. Put some weight on it and ran a mono signal path through some EQ in a DAW. Got way way way more volume generally than the stock head. Mostly in the mids and very little highs and lows playing back with a shure n44 playback stylus. Tracked and back queued ok for what that’s worth