I'll also dump my calculations here.

A brief explanation:
When you try to measure anything, you have to go through 200k in series via each data pin or directly via OUT and GND pins.
The resistor network kind of forms a ring, totaling 925 kohm.
When measuring say the 100 kohm between pins 1 and 2, it means there is also the leftover 825 kohms in parallel, and then 400 kohms in series.

That did the trick:

ftp://ftp.oldskool.org/pub/misc/temp/LPTDAC_c … al%20dream.flac
ftp://ftp.oldskool.org/pub/misc/temp/LPTDAC_c … t%20corpus.flac

That is nearly identical to the original Covox Speech Thing 😀 Lower output volume, but the clarity is an unmistakable improvement and nearly identical to the real Covox Speech Thing.

I did some informal A/B testing with the test corpus above, compared to the test corpus from the real Covox, and I was able to pick out the Covox as sounding better over 80% of the time. However, I do NOT consider that a failure! it's only because I've been doing all these tests for many evenings (I've listened to the corpus over 100 times), and am possibly biased towards how the real Covox sounds. Maybe if I had wired up 25K instead of 27K, that may have made them 100% identical, who knows. Either way, I consider this a great success.

Hopefully Serge has enough information from all these tests to decide what direction he wants to take his project. I need to move on to my own projects, but I'll be monitoring the thread.

keropi wrote:

My CVX-2s have 7.5K resistors, what is the correct value for the resistor in C1's place?

Hmm, correct to achieve what? If you want attenuation from 5Vpp to 1Vpp, the correct resistance is 7.5k/4 = 1.875k.
But in fact, most likely it will have even lower amplitude, as the 7.5k resistors load the data pins more.
So as the resistances are different, they could sound different even when recorded with same equipment and wires.

I bet it's still good enough, considering it's made of discrete resistors.

^ I have uploaded some recordings on soundcloud here: Re: CVX-2 : covox adapter , it's good but not SP good

Excellent, so we have 'cracked' the Covox-code. That 'load' resistor is what affects the sound so much.
Since it seems like no clones have managed to achieve the Covox sound and clarity, I would think that the Covox does not use a standardized resistor pack.
That 25k resistor is probably tuned specifically to get the correct load for a real IBM parallel port driving the R-2R ladder, to a line-out level (~1v).

Great combined effort here, Jepael and MobyGamer!

MobyGamer wrote:

That is nearly identical to the original Covox Speech Thing 😀 Lower output volume, but the clarity is an unmistakable improvement and nearly identical to the real Covox Speech Thing.

I did some informal A/B testing with the test corpus above, compared to the test corpus from the real Covox, and I was able to pick out the Covox as sounding better over 80% of the time. However, I do NOT consider that a failure! it's only because I've been doing all these tests for many evenings (I've listened to the corpus over 100 times), and am possibly biased towards how the real Covox sounds. Maybe if I had wired up 25K instead of 27K, that may have made them 100% identical, who knows. Either way, I consider this a great success.

I have a theory that it may be because of the discrete resistors.

I can't know what tolerance resistors it was built with, but 1% resistor tolerance is not enough when compared to a resistor network.

Resistor networks are usually available with 2% tolerance, sometimes with 1% tolerance. So the resistance value itself is not very accurate, for R=100k, the actual values are all within R=98k and R=102k. But what's more important is that the whole network in total is accurate down to 1 LSB or 0.5 LSB counts - which means that for a commonly available 8-bit network with 2% tolerance and 0.5 LSB count accuracy, the resistance values are within 0.195% of each other (0.5/256), guaranteed.

So with discrete 0.1% resistors, you can certainly build a 8-bit DAC, even almost a 9-bit DAC.

But using 1% resistors, even a 8-bit ladder is effectively a 7-bit or 6-bit DAC as it is said in this application note :
http://www.ttelectronics.com/themes/ttelectro … DERNETWORKS.pdf

@Scali:
Thanks! I don't know if this is specifically engineered or tuned value. Most likely they experimented with bog-standard resistor networks, and the 100k network is the first value crossing anyone's mind when choosing a value that's high enough not to load the data bus. And for connecting this to other audio equipment the 5V signal was too much and again the first value to cross anyone's mind is to attenuate that to 1V signal 😀 And to make manufacturing more cheaper and faster (and to avoid clones) they decided to request a custom resistor network so they don't have to solder two components...

Great results !
Sounds like the covox recordings indeed 😀
Thanks !

Jepael wrote:

So with discrete 0.1% resistors, you can certainly build a 8-bit DAC, even almost a 9-bit DAC.

For the record, the CVX-1 Serge sent me for testing used 100k 0.1% resistors (I measured them to be sure). (And I made "true covox" changes to the CVX-2 he sent me as well 😉

So for the CVX-2 with SMT 7.5k resistors:
1. Bypass R19.
2. Replace C1 with a 2k resistor.
Correct?

Consumer line level is 0.316V RMS or 0.447V Peak, so it is important to include the load resistance when measuring voltage.
If I include a 10k input impedance of a typical consumer device such as a sound-card line-in I get 0.350V peak, without it, the level is indeed 1V as MobyGamer measured.
The 100k/25k variation (as with the real Covox) is sensitive to input impedance, so if the input impedance is 1M the Covox output voltage will be 1V, but with 10k load the voltage is 0.350v.

With the 7.5k CVX-2 a 1k (or even 750ohm) resistor will be more appropriate since a 2k resistors reaches almost 1V which is almost twice the standard line level.
This variation is less sensitive to load impedance above 10k, and the output voltage will stay the same.

@MobyGamer, can you measure the Covox output again with a device (your recording computer) plugged into the Covox?

Moved to Sound.

James-F: That's why I was more focused on getting the resistances identical, so that when connected to same equipment, they sound identical. Connecting 7k5 device and 100k device to same recording equipment will have different amplitude/frequency/lowpass characteristics as the source impedance is not identical. If you bring down the voltage on a 7k5 device with about 2k load resistor, you can always add series resistance to bring up the source impedance to match original 100k device source impedance. But I still don't know if they can sound identical, as the 7k5 network will load the data outputs more, so the voltage output of the ladder itself without any extra load resistors is lower anyway. The original LPT port had 5V LSTTL chips and they have quite a high output impedance when sourcing current (it could approach 1kohm). Modern LPT ports may have push-pull 3.3V CMOS outputs so output impedance should be safely below 100 ohms, and the voltage is definitely lower.

And about the line level, there is no formal specification what is the maximum. 1Vpp audio is just as good as 2Vpp audio, but 5Vpp starts to be too much. That -10dBV (0.447Vp) is only a typical level, it does not say anything about the maximum level. Even in the next wikipedia chapter they say line outputs could reach 2Vpp (=1Vp) even when loaded with 10k. Many line outputs are capable of providing 2Vrms at full scale (5.6Vpp) and many line inputs can accept it without clipping. You can measure yourself what your sound card outputs in typical environment, say playing your favourite music when connected to home stereo line input perhaps. And even in that case, playing full scale sine wave should be much louder than music.

In broadcast devices however, the alignment levels are standardized, and there are many standards - in some broadcasting studios the +4dBu (1.736Vpeak) alignment level is the analog output level when a -20dBFS digital tone is played, so that's not the maximum either.

I have to say, it's pretty amazing just how good this sounds, considering how basic it all is.
A simple 1980s 8-bit printerport, some resistors, and a simple program that outputs one sample at a time via the timer interrupt, at 44.1 kHz, streaming from HDD.
It's not CD-quality, and it's not stereo, but quite acceptable for casual music listening.
You could convert an old 386/486 machine into a nice 'jukebox' that streams hours of music 😀

I'd say this is especially useful for laptops with no sound card.

EDIT: Now how about a Stereo-on-one project? 😁

James-F wrote:

@MobyGamer, can you measure the Covox output again with a device (your recording computer) plugged into the Covox?

My project space is dedicated to a new project that I can't easily undo at the moment. If it is vital for the project that measuring under load be done, let me know.

So I built a CVX-3 with 100k/200k ladder (using 1% 100k resistors) and 2 output variants: 25kohm load resistor and no capacitor, and no resistor but with 4.7nF capacitor. I recorded Crystal Dreams 1 demo with both setups and uploaded them here. Testing method was exactly the same as in my previous tests CVX4 : high quality covox adapter except that I had to raise input volume to 50% because of much lower level.

https://soundcloud.com/user-827772994/cvx3-10 … k-200k-25k-noc1
https://soundcloud.com/user-827772994/cvx3-10 … 200k-nor28-47nf

Both tracks are quite noisy. Perhaps it's because I have to use the mic input as the recording laptop again doesn't have a proper line input.

Scali wrote:

and it's not stereo

if the computer have two LPT ports, in theory it could play stereo music using covoxes,
so far i only know FastTracker-II that can output such 'stereo' music using two LPT ports.

Joey_sw wrote:

Scali wrote:

and it's not stereo

if the computer have two LPT ports, in theory it could play stereo music using covoxes,
so far i only know FastTracker-II that can output such 'stereo' music using two LPT ports.

In theory, even quadrophony is possible this way.
MOD Play Pro could control up to four mono DACs, if I'm not mistaken.
(See "User Defined D/A quadrophonic port" section in modplay.cfg)

That makes one for each channel in the original tracker format(s)! 😀

stamasd wrote:

Both tracks are quite noisy.

That's an understatement. Do you have access to a different recording device, to rule out the recording device as the source of distortion? Until you do that, you don't know if you made a mistake building them or not...

I will try to make another set of recordings today using a different machine, which has an actual line input. It's a new laptop that I never used for recording before, so we'll see how that goes.