Please standby ... transmission in 3.2.1...
Here we go.
The 33120 is an, now obsolete, 0.. 12 MHz Arbitrary waveform generator from HP / agilent. It's been around for a while and was their first machine in the bench instruments series. There were higher-end ARB's but this was the first affordable one due to the high integration of the electronics. It is housed in the same chassis as many other instruments form this series like the 34401 , 34970 , and others. All these ,achines come from the same 'series' in time and are now being replaced by a newer generation with fill color TFT displays.
Direct succesors are the 33250 and others.
![](http://www.siliconvalleygarage.com/eevblog/33120teardown/generator.jpg)
The above image gives an overview of the guts of this machine. One large board holds all electronics while a flatcable connects to a slave board that holds the frontpanel electronics like hte display and keypad. the frontpanel has its own local processor that handles keyboard scanning and display control. a 87c51 does all the work and drives a couple of sipex high voltage VFD driver chips. pretty boring and uninteresting. A custom built noritake VFD 14 segment (so called starburst displays) display interacts with the user.
![](http://www.siliconvalleygarage.com/eevblog/33120teardown/power.jpg)
The power section has ample heatsinking and a fan to force some airflow. the machine draws quite an amount of power becasue of all the integrated circuits present. the connectors before the caps lead to the mains transformer and an optional PLL phase lock / tcxo board (option 001 )
True to real test equipment design there is again an inguard and outguard section. The outguard is the earth-referenced i/o logic while the inguard is galvanically isolated from the mains ground.
![](http://www.siliconvalleygarage.com/eevblog/33120teardown/gpib.jpg)
The outguard logic has its own 87c51 and MP9914 GPIB controller chip.
![](http://www.siliconvalleygarage.com/eevblog/33120teardown/iso.jpg)
This section communicates over an optical link with the main processor.
![](http://www.siliconvalleygarage.com/eevblog/33120teardown/brains.jpg)
Which, in this line of equipment, is an intel 80c196 16 bit processor ( successor to the 8051 )
A custom asic, also common for this line of machinery, handles the glue logic as wel as a couple of uart's and some other logic for the 80196 system.
![](http://www.siliconvalleygarage.com/eevblog/33120teardown/rom.jpg)
Besides the asic there are only the firmware roms ( two 8 bit roms to form one 16 bit wide rom and a 32kilobyte static ram to let the firmware do it's things.
The real generator portion is formed by a DDS asic designed by HP but fabbed by TI. This custom chip has not only the direct digital synthesizer on board but has some DSP-like functionality to do signal modulation. It is hooked up to its own very fast static SRAM (cypress chip )
![](http://www.siliconvalleygarage.com/eevblog/33120teardown/dds.jpg)
Waveform data is built by the DDS chip and stored in two external fast SRAM's ( the two cypress chips )
![](http://www.siliconvalleygarage.com/eevblog/33120teardown/ramdac.jpg)
This data is scanned under control of the DDS chip , level shifted to ECL logic and sent into a very fast (for the time this was made) DAC.
![](http://www.siliconvalleygarage.com/eevblog/33120teardown/dac.jpg)
This dac is made by a company called Signal Processing Technologies (SPT) out of colorado that has been 'borged' by TI now.
The dac has current mode differential outputs and the entire post processing is done in differential mode.
The synthesized signal first hits a switchable filter to remove any unwanted spurs and artefacts from the DAC
Depending on which kind of waveform (sine/ square. Actually depends on the crest factor) ) is being created the filter is switched.
All the liitle blue components are TDK RF coils and the filter is a pure passive high-order design.
Modulation is done in the digital domain by sampling the incoming signal using an A/D and feeding the captured data to the DDS chip .
The entire signal pathway is quite large and has lots of components
![](http://www.siliconvalleygarage.com/eevblog/33120teardown/signal.jpg)
The power amplifier is constructed from discrete components and not an IC is in sight.
![](http://www.siliconvalleygarage.com/eevblog/33120teardown/amplifier.jpg)
![](http://www.siliconvalleygarage.com/eevblog/33120teardown/driver.jpg)
The last stage in the amplifier uses two big heatsinked transistors.
There is an attenuator section in the signal processing pathway as well
![](http://www.siliconvalleygarage.com/eevblog/33120teardown/atten.jpg)
Range switching is done using telecoms grade shielded dual coil relays. The amplifier output is fused on the pcb. An option is available for educational institutes to have this fuse moved to the BNC cable so it is easilty replacable is a student fries it.
This machine has been in production for a while and parts obsolecence has required a redesign of the amplifier section. The images here are form the NEW design. I will open up my other machine and take a side by side picture later. I have one of each. There are 4 transsitors which have been changed. No resistor values have changed. They went from SO-8 packages to SOT223 for some of the power transistors. The originals were Motorola designs that were killed of when motorola sold that division to On semi. The new transistors are Philips / NXP.