Emax II, Technical Description

Architecture The Emax II is a 32 voice, 16 bit, 16 stereo voice polyphonic digital sampler, based very closely on the design of the Emax and EIII, but enhanced with two custom DSP chips to replace the discrete micro controller circuitry, and to create digital rather than analog filters.

The main processor is a 16 bit National 32CG16, clocked at 20MHz (twice the speed of the EIII, but not as fast as the later EIIIx). The Emax II is entirely controlled by computer software (the operating system) which is loaded from disk at power on, into 4 DRAM’s with a total of 128 Kbytes. The small amount of OS RAM means that the software is loaded in overlay mode, which means there is a short delay as specific modules are loaded from disk each time they are used.

The scanner CPU is a Rockwell 6502 processor. The 32CG16 main processor boots the initial firmware from two 2764 EEPROM’s with a total size of 16 Kbytes.

The DSP Chips To simplify the design and to reduce costs E-mu Systems developed their own custom Digital Signal Processing chips. There are two of these. One is the main audio processor, which provides 16 channels of pitch shifting, sound memory addressing and volume scaling,. This frees up the main CPU from this core task. The other DSP implements 16 digital 4 pole filters as well as the final volume contours.

Sample Memory Up to 8 MB of sample RAM can be fitted to the Emax II stereo models, with an expansion board. The memory is either 32k or 128k byte DRAM. Different memory addressing chips are needed (PAL’s) for different sizes of memory. Memory upgrading requires new software and new PAL chips.

Support Chips Two standard 6850 UART chips provide MIDI and RS422 serial interfacing. The floppy disk gets a WD1772 controller, whilst SCSI is handled by a NCR 5380 chip. Just the same as an EIII.

Hard Disk The optional SCSI hard disk is mounted close to the motherboard and the factory drives fitted range from 40 to 127 MB. We have no information on models, so email us if you know.

Digital Sampling The Emax II samples at 16 bit resolution via a Crystal Semiconductor 5326 stereo ADC (64 times oversampling Sigma/Delta), and stores the samples into memory (and disk) as 16 bits. The Emax II replays 16 vopices via 8 channels at 18 bit resolution. It uses eight Analog Devices AD1860 DAC’s, with a 3 pole reconstruction filter over the output to smooth the “stairsteps” of the signal. The DAC’s are the same converters as on the EIIIx.
Mono Emax II’s use a Crystal Semiconductor 5126 for sampling, which has no oversampling and therefore requires an anti-aliasing filter. Both ADC’s are supplemented by digital filtering over 18kHz by the DSP chip.

The maximum sample rate of 39kHz is always used by the ADC, with the DSP doing the conversion to lower sampling frequencies. The use of the two custom E-mu VLSI chips for Digital Signal Processing has greatly reduced the chip count and hence the reliability and cost of the Emax II over the Emax 1.

Digital Filters The Emax II has 32 digital 4 pole filters, which are a very close approximation to analog filters, they even have resonance!.

OS The Operating System has minor improvements over the Emax, but it is essentially the same.

PCB’s The Emax II has a motherboard for the processors, initial memory and DSP chips, an optional memory expansion board, a front panel board for the controls and an output board. The output board for mono Emax II’s is different to the stereo models, and the motherboard is different between turbo and non-turbo.

Phase Coherence The stereo outputs of the Emax II are 1/8 of a sample out of phase. In degrees, 22.5% @ 20Khz. At 1 Khz it would be less than 1 degree. The consequence of this is a 3.2 microsecond delay between left and right stereo channels. The equivalent of having your speakers about .05 inches from being exactly in line. When Stereo is mixed back to mono, you will get an attentuation at 20 Khz of a little less than a decibel.
Thanks to the chief wizard himself !!

Weaknesses The Emax II brings 16 bit technology into a lower price point, with the use of custom DSP chips (forming the basis for future Emulators). However the memory upgrades are difficult to achieve, and the 39kHz sampling limit is a shame. The cost of memory DRAM in 1988 was still a large constraint on the design, both in turns of OS size and on sample memory. Avoid the mono Emax II as it is low on memory, and the ADC is of a slightly lower quality.
Engineering Changes
The Emax II has had a number of engineering changes – they are currently unknown.

Advertisements

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s