IndexError: tuple index out of range when assign a dask array to a dask DataFrame column

i am trying to assign a dask array to a dask dataframe but came across this weird issue.

Example:

import dask
import dask.dataframe as dd
import dask.array as da

# create a random dataframe like official documentation does
ts_df = dask.datasets.timeseries(
    '2000', '2003', freq='2H', partition_freq='2Y'
)

print(ts_df.head())

"""
                      id    name         x         y         z
timestamp                                                     
2000-12-31 00:00:00  962     Tim -0.562319 -0.331690 -0.894009
2000-12-31 02:00:00  995   Frank -0.952083 -0.725073 -1.677156
2000-12-31 04:00:00  932     Bob  0.579130  0.923146  1.502276
2000-12-31 06:00:00  997  Ursula  0.120449 -0.059130  0.061319
2000-12-31 08:00:00  975  Victor -0.947120  0.647456 -0.299664
"""

# assign a new column using dask Series works fine
ts_df['z'] = ts_df.x + ts_df.y

# da.where() returns dask.array.core.Array
new_col = da.where(
    ((ts_df.x >= 0) | (ts_df.y >= 0)),
   'N/A',
    ts_df.x.astype(str) + 'x' + ts_df.y.astype(str)
)

# assign a new column using dask array runs into error
ts_df['test'] = new_col

---------------------------------------------------------------------------
IndexError                                Traceback (most recent call last)
<ipython-input-224-392b07690d97> in <module>
      2     ((ts_df.x >= 0) | (ts_df.y >= 0)),
      3     'N/A',
----> 4     ts_df.x.astype(str) + 'x' + ts_df.y.astype(str)
      5 )

~/venvs/se/lib/python3.7/site-packages/dask/dataframe/core.py in __setitem__(self, key, value)
   3373             df = self.assign(**{k: value for k in key})
   3374         else:
-> 3375             df = self.assign(**{key: value})
   3376 
   3377         self.dask = df.dask

~/venvs/se/lib/python3.7/site-packages/dask/dataframe/core.py in assign(self, **kwargs)
   3633                         )
   3634                     )
-> 3635                 kwargs[k] = from_dask_array(v, index=self.index)
   3636 
   3637         pairs = list(sum(kwargs.items(), ()))

~/venvs/se/lib/python3.7/site-packages/dask/dataframe/io/io.py in from_dask_array(x, columns, index)
    416     dask.dataframe._Frame.to_records: Reverse conversion
    417     """
--> 418     meta = _meta_from_array(x, columns, index)
    419 
    420     if x.ndim == 2 and len(x.chunks[1]) > 1:

~/venvs/se/lib/python3.7/site-packages/dask/dataframe/io/io.py in _meta_from_array(x, columns, index)
     57         )
     58     else:
---> 59         if np.isnan(x.shape[1]):
     60             raise ValueError("Shape along axis 1 must be known")
     61         if columns is None:

IndexError: tuple index out of range

How should I assign a new column to dask dataframe correctly? if I do ts_df['test'] = tt.compute() it returns TypeError: Column assignment doesn't support type numpy.ndarray which makes sense since it's a distributed dataframe. we should still assign a distributed array to the dataframe. I tried ts_df.assign(test=new_col) also got the same IndexError.

Environments:
-- Python 3.7.6
-- dask 2.9.2