perhaps you could elaborate on the case leakage and the output level unsuitability for radio.
When testing a radio receiver it is often necessary to inject signals of less that 1uV into the antenna connector of the radio.
A signal generator will have an oscillator of some kind, and a power supply.
The RF output from the oscillator will pass through an attenuator of some kind before it goes to the output connector where we connect the screened (coaxial) cable which connects to the receiver antenna connector.
If the oscillator is not installed in a well screened box,
If the wiring to the oscillator from the power supply is not well well filtered to prevent the oscillator signal escaping from the screened box,
If the attenuator assembly is not well screened,
If the coax cable from the oscillator box to the attenuator box, and from the attenuator box to the output connector is not well screened.
Then the signal generator will radiate a signal, like a little transmitter, an we will not be able to measure the sensitivity of the radio receiver with any accuracy.
This is what I mean by "case leakage", signal just leaking out of the signal generator case.
To adjust the output level of the signal generator, we must have a well screened attenuator unit, otherwise we will not be able to make signals as small as 1uV or less.
But if you are not working on sensitive radio equipment, these features are not so important.
Look here:
I started looking for a design for a hombrew signal generator, and discovered these: http://www.ebay.co.uk/itm/New-0-40MHz-TAD9850-DDS-Signal-Generator-Module-Test-Equipment-/400503255267?pt=UK_BOI_Electrical_Test_Measurement_Equipment_ET&hash=item5d3fdab0e3 (there is a different design board on...
www.electro-tech-online.com
and here:
...to see some of my adventures in making a usable signal generator.
JimB