eTAP2HW basic dial configuration (draft)

2

Status: DRAFT open for discussion

The simple version of eTAP2 is with 3 dials and are used to fine adjust the selected unit (yes a 4th dial with 8 positions is there also to select between the emulations). In all configurations the feedback is under one position obviously pending where you would like this function. One dial is marked ‘DRY/WET’ but this is not yet final. It’s possible you need to achieve this on the outside of the kit.

The ‘SELECTOR’ requires some explanation, this selector is also a pot meter but close to the orange dots instructs the computer to select a particular configuration (as an example, all heads + head6 as feedback or heads 2,3,4 and head 5 as feedback). The choose of ‘Meazzi’s’ ABC(D)EF thus dropping the ‘D’ is based on the fact that only 5 positions are be handled and  ’D’ does not add any value to the configuration. If we look,as an example, at say selection ‘C’ the actual range for C is from dial position 4-6 everything in between is seen as a ‘C’. When automation is thought of this would become 3.3/2 volts = 1.6 volts to the computer. Using 5 area’s is therefore the best choise.

When the dial is assigned ‘SELECTOR’  without markings the actual switching requirements are not yet known in detail. The limitation is 5 positions.

All emulations are defined and worked on. There is one empty slot not defined yet. When, in the future, assuming there is an upgraded eTAP2HW called eTAP2HW+ the ‘DRY/WET’ and ‘FEEDBACK’ are taken care out outside the DSP unit. Existing dials are then re-assigned to other functions.

/eof/

A first fingerprint of a complete programmed unit

5

1/2/2012: A very first fingerprint of the young born! It’s a first programmed daughterboard with 5 emulations designed by me. This curve represents an Echomatic-I model ‘F’. The main objective was to check my EEprom programming tools and processes to be ready and fully functional when all emulations are in. Using the developer board introduces some extra hum making detailed analysis not really possible.

Seems the output level is still much too low and my tapenoise injection still too coarse. The centre frequency should be more to the right with a pivoting point around 1Khz. The sound is very acceptable with harmonics including the dry/wet ratio’s. On the right side of this curve  the drop is too quick it should be slightly extended a couple of Khz to cover the real Meazzi curve.

This set-up is also correct to carry-out extend calibrations when all emulations are in. I’ve noticed some difference between the E-1′s that should not be there. Anyway it’s a first, back to the code writing of the Roland 301!

2/2/2012: A sonogram presentation of a decaying echo sound. It’s clear that ‘high’ harmonics are fading away seems no extra shelving is required. signal/noise ratio is still (too) high and a bag filled with harmonics (lines). This is one of the tests required on each emulation to assure it really works as planned.

 

4/2/2012: Just checked the calibration of the bandpass filter for the Echomatic-II and removed -3dB on the LP side. The curve seems to be too symmetrical yet but this is the filter for the listening tests. I have the -3dB LP’s commented out so they can be put back in a seconde!

The noise floor of the unit is now -76.3 dB the noise curve as per graph. The added tape noise is below this level and set to -78dB. If an echo packet is back into the feedback loop for 3 times the added noise get noticable.

The frequency sweep through the unit reveals the feedback to be working correctly although some residue from around 5Khz is still showing a little bit above the lower frequencies. In the listening tests this will be adjusted when noticeable.

/eof/

 

eTap2HW Emulations

14

Getting to the point of the ‘final end-game ‘adding the emulations into the hardware DSP EEprom it’s time to re-address what’s going to be part of it:

  1. Echomatic I model ‘J’ modes ABCxEF
  2. Echomatic I model ‘F’ modes ABCxEF
  3. Echomatic II   ( with zero crossing distortion)
  4. VOX Longtom (with halo)
  5. PA306
  6. Roland 301 mode ’5′ with variable speed 300% range
  7. DOG (dream of glory emu derived from Meazzi)
  8. T.B.D possible Abbeyroad reverb
The Echomatics I&II are now in calibration mode, the Roland model 301 is almost ready for calibrations.  Filtering is presently revisited to assure it can be calibrated with sense, not trial and error!
In all EMU’s the tapenoise and W&F are in but on most emu’s ‘zero crossing’ distortion is taken out due to space limitations. The first daughterboard loaded with the compiled EMU’s will be asap and used for final calibrations. Before that, I’ve made arrangements to take my test-rig to Arwati who will play some Shad tunes I can use to fine adjust frequency curves. Remember that patch settings are just position’s  on the dials.
Skeleton code of both Echomatic-I and II are already offered to another ongoing project as a potential core delay model because  automation with pre-set patches of this DSP unit is rather easy.
To support general use of the DSP module as part of other projects I’ve planned the configuration offerings to an extend that it will also fit directly in all solutions.
I’ve also ordered a singleboard computer with LCD display and VU meters to be ready for an automation job try-out when the project ‘at hand’ is ready. This board features a microprocessor that includes also some DSP functionality including a LCD display and two VU meters. This unit can communicate with the FV-1 unit ‘s pot inputs and through I2C protocol with the eeprom making swapping more patches easy. The latter is outside the scope!
/eof/

Adding the Haas stereo effect ala PK for your mix

2

A write-up on a possible way of using a reverb technique resembling PeterK’s way of using reverbs and opening up your stereo field for the guitar making it a more transparent experience seems to be correct from a timing point of view as the new eTAP2 B3 is with this feature build-in:

The main idea is to add additional echo’s to the left and right of the instrument (solo guitar) but with two different timings. It’s very much like a pingpong delay. The effect applied is called the Haas Effect as the discoverer of this physical phenomenon regarding the perception of sound by the brain.

Adding those reverb’s is also adding some more body to the instrument as they are all sourced from one source so about 3dB will be added overall. Remember that the resulted sounds are not matching 100% due to the complex processed nature of adding two audio streams so it could be a little bit less.

I’m explaining this as a post processing to show all steps in sequence. Real-time is also possible and advisable to hear directly the sound.

Firstly, the main EQcurve like the smile curve or the inverted smile curve should be applied and then the echo you need for the particular song. Now you should add an extra delay with left timing 300 mS (66% Dry/wet) and a right timing 350 mS (66%dry/wet). Here you use the dry/wet ratio’s to obtain the desired results. (Yes, you could also change the 300mS to another value but make sure the delta between left and right is maintained around the 20-50mS

[Note1] the maximum should be always  < 80mS. this value is based on complex signals but for speech it should stay between 2mS and 50 mS. So, Yes 80mS is allowed but start with the low-band <50mS first!

The delayed track can differ up to +/- 10dB in volume from the original. but start with -10dB

A more advanced way is to duplicate the [source+echo] to another track and firstly apply a high-pass filter to avoid frequencies < 400Hz. Then add the extra delays with left timing 300 mS (100% wet) and a right timing 350 mS (100% wet). Here you use the mix ratio [source+echo] / [additional echo] to obtain the desired results. A good starting point of -10dB for the [additional echo] is highly recommended. This advanced approach should provide a better “wall reflections” as low frequencies are not really reflected from walls so they should stay away to avoid a muddy sound. The best way to establish the ideal setting is to mute the [source+echo] and adjust the [additional echo] to just detectable. Then adjust the [source+echo] to your liking. By using this approach the reverb is never “overdone”

[Note1] The sounds will begin to be heard as distinct when timing exceeds >50mS this is called the Haas Effect Called after Helmut Haas reference: his doctoral dissertation “The Influence of a Single Echo on the Audibility of Speech”

/eof/

eTAP2 Engineer (b3)

5

Here is the new compiled build3 of the eTAP2 family.

This engineer version is with the PP function. The performance over the eTap2 build1 is improved compatibility, performance and some required upgrades. The eTap2 engineer-pro version will be,after testing also available.

On-line help and support is embedded into the serialized unit.

/eof/

eTAP2 Engineer-pro

6

It was time to upgrade the eTAP2 family with a new VST version as  computer processors are more sophisticated these days and therefore the loading minimized for this type of VST module I’ve added the following functionality:

  • Reverb
  • Ping-Pong ala Korving (was part of build 02)
  • Input gauge

The unit is now a little bit bigger and compiled with the latest compiler to make it even more compatible. The layout of the knobs is slightly improved

The unit is in Beta mode and I would like to ask for a volunteer to check it out and work with me to test/improve the additions. The actual CPU loading on my machines is less then 5%. One of the ‘machines’ is a virtual ‘Windows XP’ desktop running on my Apple mini!! Let me know per comment if you would like to be a ‘tester’.

/eof/

How does the eTAP2 VST preset manager works?

Reply

The preset-manager, part of the eTAP2 VST module, works with JPatch data stored in the module and is therefore read-only. This data is also copied in RAM memory so any change you make to the patches (or new patches designed) are only kept in the RAM being memory that’s freed-up when the module is switched-off again.

The best way to manage this is to use the load and store all programs facility.

Let’s assume you’ve changed patch 1 and would like to move on the the next patch before saving. In this case your first patch is deleted and replaced with the default patch! Here are some simple rules so you can achieve saving them:

Rule1: Any change on a patch should be saved before moving on (switch the locked LED to unlock and just switch the save button below the LED) Then if you go to the next patch and changed it then do this procedure again…

Rule2: When all changes are in: conduct a save all programs before you’re quit using eTAP.

Rule3: When you start a new session with eTAP just load all programs with your file and it’s all there to be used.

It’s also good practice to use a naming convention for each new set of (save all programs) patches and to keep the old text files for backup. I’m using programs120108.txt for a configuration I’ve saved on 12Jan08

 

/eof/

Fingerprinting echo devices

2

A 3D frequency versus time plot of a sweep frequency as input resulting into an measurable output is a good way of establishing the character of the echo unit under investigation. In this first example I’ve added some supported lines/comments to explain how to ‘read’ these fingerprints.

On the left axis is the frequency band and on the right axis the decay (or delay time in mS). The very first part around the 0 mS is the dry part component on it’s own as there is not yet delay added (replay heads output + dry). The gradation in color of the graph is per dB value so the curve can be translated to e.a. a filter program.

let’s have a look at some basic emulations I have here in my library:

Here, firstly, is the Meazzi model Echomatic 1′ve obtained from an eTAP2 emulation based on SPICE emulations of the original Meazzi unit’s schematics.

and here the Meazzi model Echomatic2

Now, as we’re getting more aquanted with reading the 3D graphs let’s have a look at a real Meazzi tape unit with 12ax7 tubes. I’m sure this is a surprise as the frequency bandwidth of the replay heads/tape are very narrow indeed. It’s possible that this unit was a restored unit and therefore cannot be used as a reference. I’m sure the guy owning this device feels pity for the ‘electronic’ users…should he?

Next an ‘authentic’ Meazzi Trannie. That’s not up to any standard at all! Seems hum is the main contributor. just thought of sharing this also to demonstrate the tremendeous   amount of spread in the existing hardware configurations around.

I have fingerprints of more Meazzi models in the library. both the Meazzi I & II are part of the eTAP2HW version and fingerprinting are part of the sound analysis to assure it sounds correctly. Remember our ear sound-per-frequency curve is getting lower and lower.
/eof/

An advanced W&F design

6

Thought of sharing this as it’s on the shelf but possible of interest:

Based on the low-cost W&F described in this blog this design is extended to control a PT2399 delay. The unit is based on a set of 8 different oscillators  generated by a single Tiny2313 microprocessor that can be switched per requirement into 4 selected frequencies to specific parameters covering all vintage requirements.The PT2399 delay is controlled by a single transistor and could also replace the first head of a DSP delay chain ( or time subtracted from the delay line). From a cost point of view this design is very affordable while the oscillations are reliable.

The nice daughterboard with the PT2399 is designed by “AnalagCustom” as part of a 6 ‘head’ echo/reverb device here is the link to the documentation.

This design is part of a more extended echo set and kept outside the scope of the eTAP2HW deliverables.

The code used to program the Tiny2313 microprocessor is to be found here. The actual timing is changed according the study results on capstan/rollers or disc behavior as described in this blog.

/eof/

A generic sound flow using digital equipment to obtain a vintage character.

12

The requirements using equipment based on DSP and BBD devices to bring the quality into the framework of vintage equipment is firstly a 12AX7 tube. In most cases a pre-amp based on this tube is more then adequate to ‘inject’ the harmonics to your sound chain. After this stage the sound chain is divided into a ‘dry’ and a ‘wet’ part as both require specific treatment. The equalizer (qty 2) is to be adjusted per channel to achieve an optimal result.

The Echo ‘wet’ sound channel is with required delays with exact amplitudes and followed by a W&F generator based on 3 flangers/ sin oscillators (Q2/Q20 only 2). After this, a tape simulation is added to obtain that specific tape sound and compression.

The ‘Dry’ channel is directly connected to the mixer together with the complete processed ‘wet’ signal to achieve the correct sound ratio. The processed ‘Wet’ signal is, after adjusting the feedback ratio, directed to the ‘Wet’ equalizer. This Equalizer is configured as a shelving filter[*] so a particular frequency band is slightly amplified (1-2dB) generating an ever darkening sound. In most cases both equalizers are put in series so the dB/octave is achieved by minimal components count.

[*] The shelving is achieved by a bandpass and a high pass filter where the HP filter’s slope is optimized for the total centre frequency. This is also a precaution to avoid motor boating of very low frequencies

 /eof/