Simple man's tape delay using C-cassette

[jatalahd]

Hi all,

To celebrate my 99th post in this forum, I want to share my progress on the topic that I started a few months ago (https://www.diystompboxes.com/smfforum/index.php?topic=125076) on using cassette adapter tape heads in a tape echo effect.

Now I have a boxed up prototype build ready (the tape mechanism is still a stand-alone unit) and a complete schematic to share with you guys. However, I am not recommending this DIY project for anyone, because to find the suitable tape heads can be a waste of time and money, but mainly I want to share this as general info what I have done, so that others can learn from it. I recorded 4 videos on a youtube playlist, where I demonstrate my build in the main video and discuss on some topics in more detail in other videos.

The main video with the sound demo is here (I talk for the first 10 minutes, the audio demo starts at 11:00 minute mark):

The 4 video playlist is here: https://www.youtube.com/watch?v=elQLFFzyZZE&list=PLt4Qg7ko8riGNrG7ZJ8d0RwMk0DKpwM5t

The version 1.0 completed schematic is here (a 1600x1200px sized png image):


The circuit description is shortly this:
The input buffer feeds the input signal directly to the output, but also to the recording amp via a level adjustment pot. The recording amplifier is buffered with a JFET frontend, so that the following low-pass filter won't reflect to the input impedance of the recording amplifier. The high audio frequencies are boosted at the recording, and the high-frequency attenuation will be done in the playback amplifier. The output of the recording amp is taken from the emitter of Q8, where the output impedance is about 2 ohms. The series resistance following the recording amplifier output creates a current drive to the record head and also attenuates the high-frequency bias signal in the return path so that it will not affect the operation of the recording amp. Here is the frequency response graph of the recording amplifier section:



The bias oscillator is a fairly standard Clapp oscillator, providing a 44 volts rms pure sine wave in unloaded condition. Keeping that in mind, the capacitors in the oscillator section need to be rated 63V or more. When loaded with the record head and the record amplifier, the bias signal can be varied between 2 - 20 volts rms, which is just enough for proper biasing even for chrome tapes (but not metal tapes). I recommend using chrome tape over the more common ferric tape, because the chrome tape seems to behave better when recording on top of the previous recording.

The playback amplifier provides a 50 to 60 dB gain and tries to cut the high-frequencies as much as possible, but there is still hiss present due to the huge gain factor. The 6.8 nF C4 capacitor is set to resonate with the 100mH tape head, so that a low-pass resonance circuit is obtained. This provides a steep cut for high-frequencies, which need to be cut, because the huge amplitude bias signal will be present in the playback as well. The frequency response graph of the playback amplifier is here (taking into account also the tape head inductance):



The output is then a simple dry/wet signal mix using potentiometer R21.

The tape heads in my build are from a Nedis cassette adapter: https://nedis.com/en-us/product/audio/portable-audio/portable-audio-accessories/550697386/cassette-adapter-35-mm-male-black . And to obtain a suitable tape deck mechanism, one needs to disassemble some consumer-level cassette player. For example "new" Tanashin tape deck mechanisms are used in the Sony boomboxes: https://www.sony.com/electronics/boomboxes/cfd-s70, which are still sold worldwide. The deck mechanism is needed ONLY for providing a steady pinch-roller aided tape transfer to avoid wow and flutter in the recording. If this tape roller system can be done by other DIY means, then the tape deck mechanism is not required. Other "external" parts than the tape heads, C-cassette and the tape deck mechanism are not needed in this build.

I have not yet made the echo time adjustable, the stock motors in the tape decks are constant speed motors, so the speed cannot be controlled by simple voltage adjustment. I have ordered some tape deck motors from China for making tests with them, but it seems that the delivery is taking forever.

I mainly want to share this information for learning purposes. I have spent all my free time for the past 3 months working with this project, testing different approaches, tape heads and cassettes and much more and luckily the results are worth all the trouble. The main objectives were to make it run on 9 volts supply and to have a "generic" interface for "off-the-shelf" C-cassettes to be used directly in the effect. I have learned a lot during this project and hopefully the information shared here will help others to get started more easily.

[duck_arse]

jatalahd - is the connection ~R31//C16~C17 correct on your circuit diagram?

[jatalahd]

is the connection ~R31//C16~C17 correct on your circuit diagram?

Yes and no 

The tape heads I use are stereo heads and signal-wise I am using only one channel. The problem in the schematic was that it showed recording signal to be fed to right channel and the playback taken from left channel. Now that is fixed and it shows now correctly in the diagram in my earlier post.

Although only one channel is used for the effect, the schematic shows that the HF-bias is taken to both channels. This is the way I built it so in that sense it is correct, but it might not be necessary to feed the bias to the second (unused) channel. I just though that feeding the same bias to both channels in the record head prevents the magnetization of the head.

[PRR]

> same bias to both channels in the record head prevents the magnetization of the head.

No.

But if the bias was built to power two channels, and you only power one, it may put too much bias signal to the head.

[StephenGiles]

It sounds great!

[danielmeakin]

I wonder if it's possible to use a compander IC like an NE571 in order to reduce the noise AKA a dolby NR type system?

[jatalahd]

Due to further development in thread https://www.diystompboxes.com/smfforum/index.php?topic=126251.0, I am updating this thread with new schematics and sound samples. Just doing it for the sake of someone finding this old version and not the new improved one which I am now presenting here. So here is the schematic of version 2.3, which will be the last update for my project.



The major change is to use a simpler playback amplifier and a bias trap between the record amplifier and the bias oscillator. The topology of the new playback amplifier is adapted from Sony TCK81 tape deck (1980's model) and it performs much better than the one in version 1.0 schematic. The gain of the playback amplifier is still around 60 dB, but there is a dip to 50 db around 1 kHz to make more room for bass and reduce the hiss from the kilohertz range.

The other problem in the version 1.0 schematic was that the record amplifier was loading down the bias oscillator output, which caused some distortion in the recorded signal and low recording volume due to weak bias. Even when adding an untuned bias trap (notch approximately at same frequency as the bias oscillator), will improve the situation a little bit. Best results are obtained when the capacitor value in the bias trap is iterated so that maximum bias signal is reaching the recording head when the record amplifier is connected. In my setup, when the pot R32 is turned all the way down (0 resistance), I measure 44 Vrms @ 57 kHz bias signal at the tape head. Obviously that is too much, so R32 can be set accordingly to provide approximately 16 to 25 Vrms signal, depending on the tape heads and cassette type used. The cheap consumer device tape heads have relatively high inductance, so they need to be driven with higher bias. For best results in my setup I had to use that 25 Vrms bias, since I was also using a Chrome tape cassette.

Furthermore I ordered some "real" tape heads from aliexpress, and settled to use the SS15RAA4 heads. Not really quality heads, but it is an advantage since they have the tape guides in place, so the recorded tapes can be played back in any other tape player. This helped me a lot to debug the recording quality obtained from my build. I also started using a pressure pad under both tape heads, since it seemed to give better quality both in recording and playback. It was a bit troublesome to get a custom pad for the recording head, but a simple soft cushion pad that is usually put under furniture legs was enough for this purpose. Not perfect solution, but adequate.

I managed to add the motor speed control to my prototype, but after one month, the motor started choking, so it was not a good idea. I have now settled for non-changeable tape speed.

Currently I have tested three tape transport mechanisms. All have the same mechanics, but one is 15 years old, one is 35 years old and one is brand new. Here is a family picture of my "gear":



The leftmost tape transport is from new Sony boombox, the middle one is from my Grundig boombox and the right one is a slightly damaged transport, which I fixed up from broken useless old tape player. I also prepared a "stomp switch" (the red button), which simply turns the motor on and off.

Lastly, I managed to make now decent recordings for sound samples (just random playing, nothing fancy), instead of youtube videos recorded by mobile phone with crappy sound encoding. This time I used a dedicated microphone to capture the sound coming from my amplifier, so the recording is acoustic and sounds just like I hear it when playing. The recordings are 48 kHz sampled wave files and uploaded to soundcloud service, so unless soundcloud does not add any additional encoding, the recordings are closest possible to the actual sound of the echo effect.

First sample using my first tape transport mechanism, the 15 year old grundig. Here I have tuned the motor speed to maximum from the internal adjustment screw at the back of the motor. There is a 30 second "clean tone" sample at the beginning and then I start the motor to spin the tape.
https://soundcloud.com/jarmo-laehdevaara/grundig-mech

Second sample is using the new transport from Sony boombox. Here the motor is tuned slower, which causes a slight distortion to appear to high notes, but it is not that disturbing though. Again, 30 second clean tone sample in the beginning, then motor started.
https://soundcloud.com/jarmo-laehdevaara/sony-mech

Third sample is from the slightly broken tape transport. It suffers from wow and flutter. And surprisingly maybe sounds the best of all three . Clean tone sample at the beginning, then motor started.
https://soundcloud.com/jarmo-laehdevaara/hitachi-mech

I wanted to share these, because I am very happy about the end result. In the end I managed to tackle most of the noise problems, there is only slight hiss audible in the samples. But this took a lot of time and effort, so still cannot recommend this kind of project for anyone

[noisette]

Nice samples I liked the Grundig best 
and good demo guitar playing also, never boring!

I think you did great work, from the first second it sounds
more tape-y than 95% of all digital emulations (imho)

[lv1s]

Hello! Amazing work.I am very impressed.
Don't you think about using Echo-matic Tape Delay schematic as part of you project?
So, you can use your record and playback amps, and Echo-matic as delay unit.
I put picture of how it could be connected

[jatalahd]

Hello! Amazing work.I am very impressed.
Don't you think about using Echo-matic Tape Delay schematic as part of you project?
So, you can use your record and playback amps, and Echo-matic as delay unit.
I put picture of how it could be connected

I guess it would have been possible to do it like you draw it, but I am a basic BJT circuit fanatic, so I wanted to do a complete circuit with discrete components. Also my influences were old cassette recorders, where the circuits have normally been designed with discrete components. I don't think the echomatic would add much to the current design, but sure it can be admitted that the input and output terminals are not very good in my circuit. Possibly with echomatic handling the input, output and looping, the recording amp could have been designed simpler. So maybe there could be some benefits with your approach.

[Bunkey]

This is awesome man, I've wanted to build one of these for so long. The kind of project you just get lost in for months 

Really well done. Thank you for the inspiration.

[akc1973]

Magic! Cool playing too. I could see Jonny Greenwood playing with one of these in prototype form.